Why this pairing works

Ranked ideas

The ideas in brief

• In ovo: Photograph CAMs with a smartphone on a cheap macro lens. Train a free Python U-Net or use AngioTool to count vessels automatically.
• In vivo: Run a mouse Matrigel plug or rat wound, scored by haemoglobin content and CD31 staining.
• Novelty: A cheap, open-source tool that shows CAM image scores predict in vivo angiogenesis. A correlation figure (r, Bland-Altman) is the headline result.

• Scaffold: Calcine waste eggshell into hydroxyapatite and blend it with chitosan or gelatin. Add a plant osteogenic extract or a trace-ion dopant such as Sr or Zn for extra novelty.
• In ovo: Check biocompatibility and angiogenesis on the CAM.
• In vivo: Test in a rat calvarial defect, with micro-CT if you can get it.
• Why it stands out: The circular-economy story is easy to remember.

• In ovo: Inject at day 0 to 3 and score mortality, malformations, heart rate and CAM vascularity.
• In vivo: Run a dose-matched rat or zebrafish study for liver, kidney and oxidative stress markers.
• Why it stands out: Public interest is high. Adding a rescue arm (for example an antioxidant) makes it mechanistic.

• In ovo: Create a high-glucose environment that impairs angiogenesis, then test pro-angiogenic candidates (honey, curcumin, PRP, a peptide).
• In vivo: Take the best candidate into an STZ-diabetic rat excision wound.
• Why it stands out: A high-glucose CAM screen is a quick way to test diabetic-wound candidates.

• In ovo: Infect the CAM with S. aureus, Pseudomonas or Candida, then treat with essential oil, phage or an antimicrobial peptide. Score bacterial load and lesion size.
• In vivo: Use a mouse or rat infected wound.
• Why it stands out: Antimicrobial resistance is a strong theme.

• Design: Screen 8 to 10 extracts by MTT assay, then CAM, then one tumour model (Ehrlich ascites or a xenograft).
• Novelty: Report how well the in ovo ranking predicted the in vivo ranking (Spearman correlation). This is the safest and cheapest option. It becomes more novel if you choose an under-studied regional plant.

• Dressing: Electrospun or cast film containing anthocyanin from red cabbage, which changes colour with wound pH, plus an active such as curcumin.
• In ovo: HET-CAM irritation test and angiogenesis.
• In vivo: Rat wound, tracking pH against healing rate.
• Why it stands out: The colour change is visual and photographs well.

• Material: Make fluorescent carbon dots from banana peel, curry leaf or similar by hydrothermal synthesis.
• In ovo: Image them live through a window in the shell, tracking vessel uptake and tumour accumulation.
• In vivo: Run a mouse biodistribution study.
• Why it stands out: Live glowing vessels make a strong figure. It needs a fluorescence imager or UV box with a camera, which is the main cost and risk.

• Setup: A cheap LED array plus a photosensitiser such as curcumin or methylene blue.
• In ovo: Treat CAM tumours or vessels.
• In vivo: Use a mouse subcutaneous tumour.
• Why it stands out: It frames the work as affordable cancer therapy for low-resource settings.

• In ovo: Measure embryonic heart rate and rhythm by video with doxorubicin, then add a protective natural compound.
• In vivo: Rat doxorubicin model with ECG and cardiac enzymes.

• In ovo: Graft patient-derived or cell-line tumour fragments onto the CAM and test drug combinations.
• In vivo: Compare with a mouse xenograft.
• Why it stands out: The personalised-medicine angle is very strong, but it needs tissue consent and ethics approval, so it is harder. Cell lines are a safer start.

My picks

• Highest wow with low risk: Idea 2, the eggshell scaffold.
• Best for publication: Idea 1, because validation work is rare and tool-based papers get cited.
• Safest for a student timeline: Idea 6, or idea 4 for something slightly more novel.

Practical notes

• Eggs: Use fertilised eggs, incubated at 37.5°C and about 60% humidity. Use about 8 to 10 eggs per group. Plan for 20 to 30% attrition.
• Ethics: Rules for embryos differ by country and institution, so confirm with your ethics committee. Mouse and rat work needs animal ethics approval. Zebrafish larvae are often treated more leniently.
• Minimal controls: Vehicle, positive control (for example bevacizumab or sunitinib for anti-angiogenesis) and untreated.
• Statistics: One-way ANOVA with a post hoc test, plus the in ovo vs in vivo correlation.

Details

• Extract features from every CAM image: vessel density, branching points, lacunarity and fractal dimension. Combine them with dose and compound descriptors.
• Train a simple model (random forest or regression) to predict the in vivo result, such as wound closure percentage or tumour volume. Validate it on held-out compounds.
• "We predicted mouse results from an egg" is a strong headline, and it costs only your time. The risk is sample size, so you need at least 10 to 15 compounds or doses.

• In ovo: Inject probiotics, prebiotics, plant extracts or nanominerals into the egg on about day 17 to 18. This is a real poultry technique. Record hatchability and hatch weight.
• In vivo: Raise the hatched chicks for 3 to 6 weeks. Measure growth, gut villus height, microbiota, antibody titre after vaccination and heat-stress markers.
• Why it stands out: There is no rodent colony and no cross-species gap. A continuous egg-to-bird story is easy to explain, and the poultry industry may fund it.

• In ovo: Puncture a standard-size CAM vessel and apply chitosan, alum, a plant extract or gelatin sponge. Time the bleeding to stop and quantify blood loss from images.
• In vivo: Rat tail transection or liver-wound bleeding.
• Why it stands out: The cheap, standardised in ovo bleeding assay is the novelty. The CAM puncture model is much less used than the angiogenesis assay.

• In ovo: Test classical herbo-mineral preparations for embryo toxicity, malformations and CAM effects.
• In vivo: Sub-acute rat study of the safest and the most toxic preparations (liver and kidney enzymes, histology, tissue metal levels).
• Why it stands out: It addresses a real gap in the safety evidence for traditional medicines. Add metal quantification (ICP-MS or AAS) if your lab can reach it.

• Setup: A low-cost red/near-infrared LED array with a power meter.
• In ovo: Test several wavelengths and doses on CAM angiogenesis and embryo viability.
• In vivo: Rat or diabetic-rat wound, with the optimal dose from the CAM.
• Why it stands out: It finds the best dose cheaply before committing animals.

• Patch: Make dissolvable microneedles (PVA, gelatin or sugar) in a PDMS mould cast from a cheap commercial needle array or a 3D-printed master.
• In ovo: Test needle penetration and dye diffusion on the CAM. Embryo skin or CAM also works.
• In vivo: Test insulin or a drug in rats.
• Risk: The moulds and fabrication take trial and error.

• In ovo: Expose eggs to a phone or Wi-Fi router during incubation. Record hatchability, malformations, heart rate and oxidative stress.
• In vivo: Rats exposed for several weeks, with a protective antioxidant arm if you want a mechanistic angle.
• Caution: Past results in this field are mixed and often poorly controlled. Use sham exposure, a measured field strength and blinded scoring.

• In ovo: Measure venom-induced haemorrhage in embryos and test plant extracts as inhibitors.
• In vivo: Mouse neutralisation assays.
• Why it stands out: Embryo screening could cut the number of mice used in venom neutralisation work.
• Risk: Venom is hard to obtain and regulated, so I rated it low on feasibility.

Combinations for the most impact

• Eggshell hydroxyapatite scaffold (from the first list) + AI vessel quantification + digital twin (A). The story is egg waste to a bone graft, assessed by an open-source tool and predicted by a model.
• Hyperglycaemic CAM + LED photobiomodulation (E) + digital twin (A). This is an affordable diabetic-wound platform, with the dose chosen in ovo.
• Haemostat (C) + smart dressing (from the first list). A first-aid kit concept.

Next step

Details

• In ovo: Film the embryo through a window or by candling with a phone. Extract heart rate and motion with free signal-processing code (Eulerian video magnification or simple intensity tracking). Use it as a non-invasive, repeatable toxicity or cardiotoxicity marker over days.
• In vivo: Compare with rat ECG after the same drug (doxorubicin, a QT-prolonging drug or an energy drink).
• Why it stands out: You replace expensive instruments with a phone and open code. The key step is validating it against a reference method.

• In ovo: Expose embryos or the CAM to UV-B or UV-C through a film or cream. Score survival, vessel damage and oxidative markers.
• In vivo: Mouse or rat UV skin model (erythema, thickness, MDA and histology).
• Extra: Measure SPF in vitro by spectrophotometer. A UV lamp with a measured dose is the main equipment need.

• In ovo: Grow tumour spheroids, 3D-printed or hydrogel scaffolds, or stem cell constructs on the CAM for several days so chick vessels grow in.
• In vivo: Implant the prevascularised construct in a mouse or rat and test whether it survives better than an unvascularised one.
• Why it stands out: It is a creative use of the egg as a cheap incubator for tissue engineering. The main risk is that the vessels are avian, so check for host integration and rejection.

• In ovo: Graft endometrial cells or tissue and test anti-angiogenic or anti-adhesion agents. Ovarian cancer metastasis on the CAM is well described, but verify the endometriosis protocols in the literature before committing.
• In vivo: Mouse endometriosis model.
• Why it stands out: Women's health is under-funded and reviewers like the topic. Use cell lines first, because human tissue needs ethics approval and consent.

• In ovo: Induce hypoxia with a low-oxygen setup or chemically (for example CoCl2). Test protective agents such as rhodiola, ginkgo or antioxidants on survival, vascular density and HIF-1α.
• In vivo: Rat hypobaric chamber if available, or a chemical hypoxia model (sodium nitrite).

• In ovo: Treat with valproate or another teratogen, then add folic acid, choline or a plant antioxidant. Score neural tube defects, limb and heart malformations, and size.
• In vivo: Pregnant rat with the best rescue dose.
• Why it stands out: The rescue design gives you a mechanistic story. Pregnant-animal work has stricter ethics and a longer timeline.

• In ovo: A CAM inflammation model is cheap, so screen many extracts.
• In vivo: Carrageenan rat paw oedema.
• Note: This is the easiest and safest option, but the least exciting. It works well as a base for idea 1 from the previous list, the AI quantification.

• In ovo: Treat with lead, cadmium or arsenic and test moringa, chlorella or coriander extract.
• In vivo: Rat model with blood and tissue metal levels and liver and kidney markers.

Ways to raise the wow factor in any project

• Time-lapse imaging. A phone on a clamp or a 3D-printed holder can film CAM vessel growth over 24 to 72 hours.
• Open data and code. Posting images and scripts (for example on Zenodo or GitHub) raises citation chances.
• Preregistration of your design and endpoints, which reviewers value.
• A single visual summary. A clear egg-to-animal workflow figure and one striking before and after image.
• A frugal comparison. Show cost per experiment against a mouse-only design.
• Negative controls and blinding. Score images blinded, which adds credibility at no cost.

Next step

1. Open-source Frugal Egg Lab kit (9.0)

• Concept: A complete, cheap, reproducible kit for CAM experiments. It has a 3D-printed or cardboard imaging box with a phone clamp and LED ring, a standard window-cutting and sealing protocol, and free software that counts vessels.
• In ovo: Run an anti-angiogenesis and a pro-angiogenesis screen with known controls (bevacizumab or sunitinib, and VEGF or a pro-angiogenic extract).
• In vivo: Validate against one mouse Matrigel plug or rat wound experiment, scored by haemoglobin and CD31.
• Why it scores high: Labs with no imaging equipment can use it, and it is publishable as a methods paper.
• What holds it back: Image consistency across labs. To get to 10, test the kit in a second lab and report inter-lab agreement.

2. Public benchmark plus digital twin (8.75)

• Concept: Build the first small open benchmark of CAM results for 15 to 20 compounds whose in vivo potency is already published (known anti-angiogenic and pro-angiogenic drugs). Then test whether a simple model predicts the published in vivo ranking.
• In ovo: Dose-response of each compound on the CAM.
• In vivo: Mostly literature data, plus one small confirmatory experiment with a new compound the model never saw.
• Why it scores high: Almost no animals are needed, and a dataset others can reuse is a durable contribution.
• What holds it back: Published in vivo data are inconsistent across species and doses. To reach 10, preregister the prediction and test it on the unseen compound.

3. Egg-to-bird feeding of waste-derived bioactives (8.75)

• Concept: Inject waste-derived bioactives (eggshell calcium nanoparticles, fruit-peel polyphenols or similar) into the egg at about day 17 to 18, then raise the chicks.
• In ovo: Hatchability, hatch weight and embryo toxicity.
• In vivo: The same birds for 3 to 6 weeks: growth, gut villus height, antibody response to vaccination and heat-stress markers.
• Why it scores high: There is no rodent colony, the cost is the lowest of all seven, and the poultry industry may support it.
• What holds it back: Novelty is moderate because in ovo feeding is established. To reach 10, pair a new waste-derived compound with a measurable health outcome such as vaccine response.

4. Everyday exposome in an egg (8.5)

• Concept: Test substances people use daily: tea-bag microplastic leachate, instant-noodle cup leachate, vape liquid, energy drinks and hair-dye ingredients. Compare them on one standard embryo scoring system.
• In ovo: Mortality, malformations, heart rate (from video) and CAM vascularity.
• In vivo: Rat or zebrafish for the one or two worst offenders, with liver, kidney and oxidative stress endpoints.
• Why it scores high: Everyone can relate to it, and press and social media will pick it up.
• What holds it back: Real-world dose relevance. Dose-match to realistic human exposure to avoid overclaiming.

5. Eggshell-to-bone with AI scoring (8.5)

• Concept: Calcine eggshell waste into hydroxyapatite and blend it with chitosan or gelatin, with or without a plant osteogenic extract.
• In ovo: Biocompatibility and angiogenesis on the CAM, scored with the idea 1 pipeline.
• In vivo: Rat calvarial defect with micro-CT or histology.
• Why it scores high: The "egg to bone" story gives the strongest visual and narrative of all seven.
• What holds it back: Eggshell hydroxyapatite is known, and imaging adds cost. To reach 10, add a clear twist, such as a trace-ion dopant, and use the CAM score to select the formulation.

6. Kitchen-to-clinic haemostat (8.5)

• Concept: Compare cheap haemostats: chitosan from shrimp-shell waste, alum, turmeric, gelatin sponge and honey.
• In ovo: Puncture a standard CAM vessel and time the bleeding to stop. Quantify blood loss from images.
• In vivo: Rat tail transection or liver-wound bleeding.
• Why it scores high: Very low cost, a clear practical use in low-resource first aid, and an easy outcome to measure.
• What holds it back: Standardising the puncture. To reach 10, use a fixed needle gauge, a template and blinded scoring.

7. Diabetic-wound platform, stacked (8.0)

• Concept: Combine a hyperglycaemic CAM, LED photobiomodulation dose-finding and a pH-indicator dressing.
• In ovo: Screen dressing candidates and LED doses under high glucose.
• In vivo: Diabetic rat wound (STZ), tracking healing rate and wound pH.
• Why it scores high: It is a cohesive story with several outputs.
• What holds it back: It needs more skills (dressing fabrication, LED dosimetry and the diabetic model) and a larger scope, so it is better as a PhD plan than a short project.

My recommendation

• AngioTool is a free NIH tool for vessel length, density, branching and lacunarity (Zudaire et al., PLoS One 2011).
• A commercial IKOSA CAM app does deep-learning vessel segmentation (Kuri et al., Cells 2022).
• An AI-based CAM angiogenesis method was published in 2022 (Faihs et al.).

What it is

1. A standard imaging box. It is cardboard or 3D-printed, with a matte white interior, a diffuse LED ring, a fixed phone holder at a fixed height, and a ruler and colour card in every frame.
2. A standard CAM protocol. It covers eggs, windowing, how the test substance is applied, timing and scoring rules.
3. Free analysis software. A Python or ImageJ pipeline gives vessel area %, total length, branch points and lacunarity as a CSV file.
4. Validation. You compare it with manual counting, AngioTool and one in vivo experiment.

Workflow

1. Incubate fertilised eggs at about 37.5°C and about 60% humidity.
2. On about day 3, cut a small window, or use an ex-ovo shell-less culture.
3. On about day 7 to 8, place a filter-paper disc, silicone ring or gelatin sponge with the test substance on the CAM.
4. Photograph at about day 10 to 12 in the box. Fix the settings and take a photo of the colour card each time.
5. Count only eggs that survive. Report attrition.

• Vehicle control.
• Pro-angiogenic control (bFGF or VEGF).
• Anti-angiogenic control (sunitinib or suramin).

• Start with the green channel, where blood vessels show best contrast.
• Baseline: a Frangi vessel filter with thresholding.
• Optional: train a small U-Net on 30 to 50 hand-annotated images.
• Outputs: vessel area %, total length, branch points and lacunarity.

• Mouse Matrigel plug: Measure haemoglobin (Drabkin) and CD31 staining. This is the standard comparison for angiogenesis.
• Rat excision wound: Use it if you are testing a healing compound. Measure wound closure and granulation tissue.

Example design (my suggestion)

• In ovo: 3 test compounds x 3 doses + 3 controls = 12 groups, with 8 to 10 eggs per group. That is about 100 eggs, so plan for about 130 with attrition.
• In vivo: 4 groups (vehicle, positive control, two doses of the best candidate) with 6 mice per group, or about 24 animals. Run a formal power calculation once you have pilot variance.

Validation steps

Timeline (about 12 weeks)

Cost

Outputs

• A methods paper or short report.
• An open GitHub repository with the box design, code and a step-by-step protocol.
• A shared image dataset with annotations.
• A ready-made teaching tool for other labs.

Risks

• Lighting and glare are the biggest threat to reproducibility. Use polarising film and always include the colour card.
• Egg variability. Source eggs from one supplier and use fixed incubation conditions.
• Annotation effort. Manual labelling takes time. Start with the Frangi baseline and only add the U-Net if it improves agreement.
• Ethics. Embryo rules vary by country, and mouse or rat work needs animal ethics approval.
• Novelty pushback. Cite AngioTool and IKOSA, and state clearly that your contribution is the open, validated, low-cost kit.

Next step

The mouse model I recommend: the Matrigel plug assay

How to do it

1. Pick the test set from the egg results. Choose compounds that cover the CAM range: one strong effect, one moderate effect and one with no effect. Add a vehicle control and a known positive control. This is what lets you test whether the egg ranking predicts the mouse ranking.
2. Preregister the prediction. Write down the expected order before the mouse study begins.
3. Prepare the plugs. Mix cold liquid Matrigel with heparin and an angiogenic factor (bFGF or VEGF), plus the test compound or vehicle. For a pro-angiogenic compound, add it alone and compare with the vehicle plug.
4. Inject the mixture subcutaneously (for example the flank) with a cold syringe. It gels at body temperature.
5. Harvest at about day 7 to 14 (use the timing in your chosen protocol).
6. Read out with:
   • Photographs in the egg-lab box (colour and redness).
   • Haemoglobin content (Drabkin method), the cheap quantitative readout.
   • CD31 immunostaining if you have the antibody, as the histology confirmation.

Group sizes

Options if you cannot use Matrigel

• Matrigel cost. It can be the largest single expense, so get a local quote early. Dilute it only as the protocol allows.
• Directed in vivo angiogenesis assay (DIVAA). The silicone-tube kits are also available, but the plug is cheaper and simpler.
• Mouse excision wound. Use this if your compounds are wound-healing agents. It needs no Matrigel and is cheaper. Measure closure, granulation tissue and hydroxyproline.
• Ehrlich ascites or solid tumour in Swiss albino mice. Use this if your compounds are anticancer. Measure tumour volume, survival and vessel density. It is widely used in low-resource labs, but it measures anticancer effect more than angiogenesis.

Analysis plan

• Compare groups with one-way ANOVA and a post hoc test, or Kruskal-Wallis if the data are skewed.
• Test whether the CAM ranking matches the mouse ranking with a Spearman correlation or simple rank agreement.
• With only 3 to 4 compounds, a correlation coefficient is weak. Add doses (for example 2 doses of each compound), report effect sizes with confidence intervals, and describe the correlation as exploratory.
• Report the full results, including cases where the egg and the mouse disagree.

Mouse study timeline (about 4 weeks)

Ethics and approval

• Get approval from your institutional animal ethics committee before you start (in India this is the IAEC under CPCSEA). Rules differ by country, so check yours.
• State the 3Rs argument: the egg screen reduces the number of compounds tested in mice, so fewer animals are used.
• Set humane endpoints and keep the procedure short. This model involves a small subcutaneous injection and a short experiment.

Updated project summary

1. Kombucha-cellulose "living" wound dressing (8.5)

• Concept: The cellulose pellicle that grows on kombucha (the SCOBY) is a cheap bacterial cellulose sheet. Purify it, load it with a plant bioactive such as curcumin, and optionally add a pH-indicator dye (anthocyanin) so it changes colour as the wound pH shifts.
• In ovo: CAM biocompatibility, irritation and angiogenesis.
• Mouse: Full-thickness excision wound (splinted if available). Measure closure rate, histology and wound pH.
• Why it works: The material costs almost nothing and the story is memorable ("tea fungus to skin patch").

2. Frugal drug repurposing for anti-angiogenesis (8.25)

• Concept: Screen 15 to 20 cheap, off-patent drugs (examples of candidates: itraconazole, mebendazole, doxycycline, propranolol, metformin, disulfiram) on the CAM. Pick the best for mouse testing.
• In ovo: Dose-response, scored with the Frugal Egg Lab kit.
• Mouse: Matrigel plug, or Ehrlich or B16 tumour if you want an anticancer angle.
• Why it works: Drugs with known safety profiles are easier to take forward, and it answers a practical question for low-resource settings. Check that the mouse dose is a sensible human-equivalent dose.

3. Sentinel egg for food adulterants and cosmetics (8.25)

• Concept: Use embryos as a cheap toxicity screen for things consumers worry about. Examples: metanil yellow in spices, formalin in fish, carbide-ripened fruit extracts, skin-lightening creams, kajal. Adapt the list to your country.
• In ovo: Embryo survival, malformations, heart rate by phone video (see the earlier video-readout idea) and CAM vascularity.
• Mouse: Short sub-acute oral or dermal study of the worst one or two (liver and kidney enzymes, histology).
• Why it works: Direct public relevance and strong press potential. Use realistic exposure doses, to avoid alarming conclusions.

4. Whole-egg dressing (8.25)

• Concept: Use all the egg: an eggshell membrane sheet, an egg-white (ovalbumin) hydrogel, and calcium from the shell. Compare them alone and combined.
• In ovo: CAM angiogenesis and biocompatibility.
• Mouse: Excision wound.
• Why it works: The material and the test model both come from the egg. Eggshell membrane is a known biomaterial, so your novelty is the combination and the CAM-to-mouse validation.

5. Phage cocktail from local sewage (7.75)

• Concept: Isolate bacteriophages from sewage or river water that kill wound pathogens (for example Pseudomonas aeruginosa, S. aureus) and formulate them in a gel.
• In ovo: CAM infection model with treatment. Score lesion size, bacterial load and embryo survival.
• Mouse: Infected wound model.
• Why it works: Antimicrobial resistance is a strong theme and the raw material is free.
• Risks: You need a microbiology lab and appropriate biosafety approval for the pathogens. Phage isolation takes skill.

6. Smokeless tobacco and areca nut toxicity (7.75)

• Concept: Compare extracts of regional products (gutka, khaini, areca nut, betel quid) at realistic doses.
• In ovo: Embryo toxicity, CAM vascular effects and oxidative stress markers.
• Mouse: Oral mucosa or cheek exposure over several weeks, with histology and a protective antioxidant arm if you want a mechanism.
• Why it works: Strong local public health relevance. Long exposure studies take time, so plan the timeline.

7. Regional honey passport (7.75)

• Concept: Collect honeys from different regions or floral sources and compare them: phenolics, antioxidants and angiogenic activity.
• In ovo: CAM dose-response.
• Mouse: Excision wound with the best and worst honey.
• Why it works: Cheap, easy and locally meaningful. Make it more novel by profiling the chemistry (total phenolics, a few marker compounds) and linking it to the wound result.

8. Edible plant nanovesicles (7.5)

• Concept: Isolate nanosized vesicles from ginger, grape, wheatgrass or lemon juice. Test them on the CAM for angiogenesis and tissue effects.
• Mouse: Wound healing or a colitis model.
• Why it works: This is a fashionable research area with a strong wow factor.
• Risks: Isolation by ultracentrifugation needs equipment. Precipitation-based isolation is cheaper but less pure, so plan to characterise your vesicles (size by DLS or electron microscopy).

My picks

• Best overall: Idea 1, the kombucha dressing. It is cheap, visual and simple to run, and the mouse wound model is well established.
• Best for public interest: Idea 3, the sentinel egg.
• Best fit with the Frugal Egg Lab kit: Idea 2, drug repurposing, because the egg screen and mouse plug are the project.

• Novelty: how new the combination is. I cap it at 9.
• Cost: 10 means cheapest.
• Feasibility: can one student or lab finish it in 3 to 4 months.
• Wow: visual impact and story appeal.

Topics and ratings

1. Kombucha cellulose dressing (9.25)

• In ovo: CAM biocompatibility, irritation and angiogenesis, to choose the best curcumin and dye loading.
• Mouse: Excision wound with groups for untreated, plain cellulose, commercial dressing and the loaded dressing. Measure closure rate, histology and wound pH.
• Scores:
  • Novelty 8: bacterial cellulose dressings exist, but the pH indicator plus CAM-guided optimisation is a fresh combination.
  • Cost 10: the material is a by-product of kombucha.
  • Feasibility 9: simple chemistry and a standard wound model.
  • Wow 10: the dressing changes colour as the wound changes, and it photographs well.
• To raise novelty to 9: show that the colour change predicts healing outcome.

2. Sentinel egg for food adulterants (9.0)

• In ovo: Embryo survival, malformations and heart rate from phone video, for 5 to 8 locally relevant substances at realistic doses.
• Mouse: Sub-acute oral or dermal exposure to the worst one or two (liver and kidney enzymes, histology).
• Scores:
  • Novelty 8: embryo toxicity testing is known, but the phone-based readout and the local adulterant focus are new.
  • Cost 9: eggs and a phone.
  • Feasibility 9: short experiments.
  • Wow 10: it speaks directly to what people eat and use.
• To raise novelty to 9: validate against a reference substance panel with known toxicity.

3. Open egg-to-mouse angiogenesis platform (9.0)

• In ovo: CAM screen of 10 to 15 compounds with the Frugal Egg Lab kit.
• Mouse: Matrigel plug with vehicle, positive control and three compounds spanning the CAM range. Measure haemoglobin and CD31.
• Scores:
  • Novelty 8: tools like AngioTool and IKOSA exist, so the novelty is the open, validated, low-cost kit.
  • Cost 10: almost no equipment.
  • Feasibility 9: well-established assays.
  • Wow 9: strong methods and data story.
• To raise novelty to 9: test the kit in a second lab.

4. CAM-based digital twin for wound dressings (9.0)

• In ovo: CAM screen of several dressing formulations, with image features extracted by the kit.
• Mouse: Excision wound for a subset of formulations. Test whether the model, trained on the CAM data, predicted the healing ranking.
• Scores:
  • Novelty 9: predicting mouse outcome from egg images is uncommon.
  • Cost 9: the cost is mostly the mouse study.
  • Feasibility 8: it needs enough formulations (at least 8 to 10) for the model.
  • Wow 10: "we predicted mouse healing from an egg."
• To raise feasibility to 9: use simple models, preregister the prediction and keep the formulation set small but varied.

5. Egg-waste hydroxyapatite for bone (8.75)

• In ovo: CAM angiogenesis and integration of scaffold variants.
• Mouse: Critical-size calvarial defect, assessed by micro-CT if available and histology.
• Scores:
  • Novelty 8: eggshell hydroxyapatite is known, so use a twist such as a trace-ion dopant.
  • Cost 9: waste material, but imaging adds cost.
  • Feasibility 8: surgery and imaging are more demanding.
  • Wow 10: "egg to bone."

What I recommend

Novelty correction

• Kombucha-derived cellulose wound dressings have been published (Dey et al., 2024).
• Anthocyanin-loaded bacterial cellulose pH indicators exist (Zubova et al., 2024; Melnyk et al., 2025).

Title

Aims

1. Make and characterise kombucha cellulose dressings with different curcumin and anthocyanin loadings.
2. Screen them on the chick CAM for irritation and angiogenesis.
3. Test selected formulations in a mouse excisional wound model.
4. Test whether the CAM data predicted the mouse ranking, and whether dressing colour tracks wound pH and healing.

1. Dressing preparation (weeks 1-3)

1. Grow the pellicle. Use a kombucha culture in sweetened tea at room temperature, static, for about 1 to 2 weeks.
2. Purify. Wash, treat with dilute NaOH with heat to remove cells, then rinse to neutral pH. Sterilise by autoclave or another method that does not damage the dye.
3. Load the actives. Soak in curcumin (dissolved in a small amount of solvent) and/or red-cabbage anthocyanin extract.
4. Characterise. Record weight, thickness and water uptake. Measure release into PBS over 24 hours by UV-Vis. Photograph the colour at known buffer pH values (for example 4 to 9) to build a calibration curve.

2. In ovo screen (weeks 3-6)

• Groups: 8 formulations + untreated + a commercial dressing + a pro-angiogenic control (bFGF or VEGF disc) = 11 groups.
• Eggs: 10 per group is 110, so order about 140 with attrition.
• Method: Window the egg early. Apply a disc-sized piece of dressing on the CAM at about day 7 to 8. Photograph at about 48 to 72 hours in the egg-lab imaging box.
• Irritation: Run a separate HET-CAM test on day 9 to 10 eggs (score haemorrhage, vessel lysis and coagulation over a few minutes).
• Readouts: Vessel area %, total length, branch points, lacunarity, avascular zone, irritation score and embryo survival.
• Analysis: Blinded scoring, ANOVA or Kruskal-Wallis, and report all groups, including failures.

3. Mouse experiment (weeks 6-10)

• Wound photographs at days 0, 3, 5, 7, 10 and 14, with a ruler and colour card in the egg-lab box.
• Dressing colour (hue or RGB) at each time point.
• Wound surface pH if you have a flat pH probe or strips.
• Histology at day 14: H&E for re-epithelialisation and granulation thickness, CD31 for vessel density.

4. Digital-twin analysis (Topic 4)

• Inputs: CAM features per formulation (vessel area %, branching, lacunarity, irritation score) plus the curcumin and anthocyanin levels.
• Target: Mean mouse wound closure per formulation.
• Models: Keep them simple (ridge regression or random forest) with two or three features chosen in advance. Use leave-one-formulation-out cross-validation.
• Benchmarks: Compare against predicting the average, and against a model using only curcumin dose. If the CAM features do not add accuracy over curcumin dose, say so.
• Reporting: Spearman rank correlation and mean absolute error.
• Honest limit: With only 4 to 6 mouse formulations, treat this as exploratory. Preregister the predictions before the mouse study unblinds.
• Colour analysis: Correlate dressing hue with surface pH and with closure.

Timeline

Cost

Risks and fixes

• Contamination of the pellicle: Culture in clean vessels, sterilise before use.
• Batch variability: Grow all dressing pieces from the same batches and record their thickness.
• Colour drift: Fix lighting and use the colour card in every photograph.
• Few formulations for the model: Add more formulations to the CAM screen even if only some go to the mouse.
• Dressing adherence in mice: Use the standard secondary cover and check dressing position daily.

Expected outputs

• A dressing recipe and its characterisation.
• A CAM dataset with images, features and scores.
• A mouse healing dataset.
• A prediction analysis and open code.

Draft abstract (to fill with your results)

1. Sentinel egg for food adulterants (9.0)

• In ovo: Test 5 to 8 locally relevant substances at realistic doses. Record survival, malformations and heart rate from phone video.
• Mouse: Sub-acute oral or dermal dosing of the worst one or two. Measure liver and kidney enzymes and histology.

2. Embryo heart-rate video screen for cardioprotection (8.0)

• In ovo: Treat embryos with doxorubicin, with and without plant compounds, and film the heart. Chick embryos have already been used to detect drug cardiotoxicity (for example trastuzumab in chick embryos), so the novelty is the cheap video readout and the protectant screen.
• Mouse: Doxorubicin cardiotoxicity with the top candidate. Measure ECG or echo if available, troponin or CK-MB, and cardiac histology.

3. Egg-grown oncolytic Newcastle disease virus (7.75)

• In ovo: The egg is the cheap way to grow the virus. Then test its effect on tumour cells grafted on the CAM.
• Mouse: Syngeneic or xenograft tumour with intratumoral virus.
• Why it stands out: The egg does two jobs, and oncolytic virus therapy is an exciting area. My PubMed search only returned old virology papers on NDV growth in the CAM, so I could not confirm how novel the CAM tumour angle is.
• Risks: NDV is a poultry pathogen. Use a mild vaccine-type strain, get biosafety approval and contain it away from any birds. This is why feasibility is only 6.

4. Drug-combination synergy on CAM tumours (7.75)

• In ovo: Graft tumour cells and test 2-drug combinations at several doses. Compare tumour size or weight.
• Mouse: Ehrlich tumour or a syngeneic tumour with the best combination against each drug alone.
• Why it stands out: Combination screening is expensive in mice and cheap in eggs. Calculate synergy with a standard method (Bliss or Loewe).

5. Antithrombotic natural product screen (7.5)

• In ovo: Induce a small vessel injury on the CAM and measure clot formation or flow. My search did not find an established CAM thrombosis model, so you would need to develop and validate it. That gives novelty but adds risk.
• Mouse: Tail bleeding time (to check for excess bleeding) plus collagen/epinephrine-induced pulmonary thromboembolism survival.

6. Herbal UV photoprotection (7.5)

• In ovo: Expose embryos or CAM to a measured UV dose through a thin film of formulation. Score survival, vessel damage and oxidative markers.
• Mouse: UVB skin exposure. Measure erythema, skin thickness and MDA, plus histology.
• Needs: A UV lamp and a UV meter.

7. High-glucose embryo to diabetic mouse (7.0)

• In ovo: Inject glucose to create a high-glucose environment and test protective compounds on vessel and embryo outcomes.
• Mouse: STZ diabetic mice. Measure retinal or renal vascular leakage (for example Evans blue) and kidney histology. This model needs weeks, so plan the timeline.

8. Anti-metastasis CAM screen (6.75)

• In ovo: CAM metastasis methods are established (Maacha and Saule) and a 2026 paper reports digital PCR for quantification, but digital PCR is expensive. Regular qPCR for human DNA is the cheaper route.
• Mouse: B16F10 tail-vein injection and lung nodule counts.
• Why it ranks last: Novelty is low because the method is established.

My picks

• Highest overall: Idea 1, the sentinel egg.
• Best balance of safety and feasibility: Idea 2, the cardioprotection screen.
• Biggest wow: Idea 3, the oncolytic virus, if you can handle the biosafety requirements.

1. Anthracycline cardiomyopathy (8.0)

• In ovo: Treat embryos with doxorubicin with or without candidate compounds. Film the heart with a phone and extract heart rate and rhythm. Chick embryos have been used for cardiotoxicity testing before (trastuzumab example), so your novelty is the cheap video readout and a protectant screen.
• Mouse: Doxorubicin cardiotoxicity regimen from a published protocol, with the best candidate. Measure ECG or echo if available, troponin or CK-MB, and cardiac histology.
• Why it ranks first: Short experiments and clear endpoints.

2. Psoriasis (7.75)

• In ovo: CAM vessel and irritation scoring to rank candidates.
• Mouse: Topical imiquimod on shaved back skin of BALB/c or C57BL/6 mice for about a week. Score erythema, scaling and thickness (a PASI-like score) and examine epidermal thickening on histology. The model is short and cheap (example study).
• Caveat: My search found no established CAM psoriasis model. The CAM only captures the angiogenic and irritant components of the disease, so state that limit clearly.

3. Diabetic retinopathy (7.25)

• In ovo: There is precedent for hyperglycaemia-induced angiogenesis defects in the CAM (Larger et al., Diabetes 2004) and for a CAM/human-vitreous platform in diabetic retinopathy (Rezzola et al., 2020).
• Mouse: STZ-induced diabetes in C57BL/6 mice, with retinal vascular leakage (Evans blue) or acellular capillary counts on retinal flat-mounts.
• Risk: The diabetic phase takes several weeks, so the timeline is longer.

4. Oral squamous cell carcinoma (7.25)

• In ovo: CAM tumour growth and vessel assays. CAM has even been tried in the clinical workflow with patient tissue (Kauffmann et al., 2018). If you use patient samples, you need ethics approval and consent.
• Mouse: Cell-line xenograft in immunodeficient mice (shorter but costly) or the 4NQO carcinogenesis model (cheaper but takes months).
• Why it stands out: Strong relevance where tobacco and areca nut use is common.

5. Endometriosis (7.0)

• In ovo: Human endometrial fragments form endometriosis-like lesions on the CAM (Maas et al., 2001), and angiostatic agents and MMP inhibitors reduced lesions there (Nap et al., 2005). The model is established, so novelty comes from what you test.
• Mouse: Syngeneic uterine-tissue transplant into the peritoneum. Measure lesion size, vascularity and histology.
• Note: Human tissue needs consent and approval.

6. Candidiasis (7.0)

• In ovo: Infect the CAM or embryo and score survival and fungal burden. I did not verify the specific published CAM Candida protocol, so check the literature first.
• Mouse: Oral or vaginal candidiasis, with fungal CFU counts and histology.
• Needs: A microbiology lab and appropriate biosafety approval.

7. Triple-negative breast cancer (6.75)

• In ovo: MDA-MB-231 tumours on the CAM, treated with drug pairs (checkerboard design).
• Mouse: 4T1 syngeneic tumour in BALB/c mice. Track tumour growth and lung metastasis.
• Why it ranks lower: CAM breast tumour work is well established, so novelty depends on your drug combinations.

8. Rheumatoid arthritis (6.5)

• In ovo: CAM has been used to compare anti-angiogenic effects in RA and osteoarthritis synovium (Ribatti et al., 2000).
• Mouse: Collagen-induced arthritis in DBA/1 mice (more costly) or a cheaper adjuvant-induced model. Score paw swelling, clinical arthritis score and joint histology.

9. Retinopathy of prematurity (6.5)

• In ovo: CAM angiogenesis screen of candidates.
• Mouse: Neonatal pups exposed to high oxygen, then room air, with retinal flat-mount to measure neovascular tufts and avascular area.
• Risk: It needs an oxygen chamber and neonatal mouse handling.

My picks

• Easiest and fastest: Idea 1 (anthracycline cardiomyopathy) or idea 2 (psoriasis).
• Best story: Idea 4 (oral cancer), if you can handle the longer mouse timeline.
• Established models, so low risk: Ideas 5 and 7.

Best topic: doxorubicin cardiomyopathy

Novelty correction

• Smartphone measurement of chick embryo heart rate (Phuphanin et al., 2019).
• In ovo video of the beating heart used to study antiarrhythmic drug effects (Kockova et al., 2013).
• A chick embryo toxicity test compared with rodents for 20 drugs (Nishigori et al., 1992).

The key idea (the wow factor)

• Measure the embryo's heart rate and rhythm (cardiac effect).
• Graft tumour cells on the CAM of the same eggs and measure tumour size (anti-cancer effect).

Design

• Groups: vehicle, doxorubicin alone, doxorubicin + dexrazoxane (the approved cardioprotectant, as the positive control), and doxorubicin + 4 to 6 candidates (for example curcumin, quercetin, resveratrol or a local plant extract). Add candidate-alone groups for toxicity. Use about 10 eggs per group.
• Dose-finding pilot: Find a doxorubicin dose that lowers heart rate or survival without killing most embryos.
• Readouts: Heart rate by phone video, beat-to-beat variability, survival, malformation score and, if possible, an oxidative stress marker (MDA) in embryo tissue.
• Tumour check: Graft tumour cells on the CAM in a parallel set of eggs. Measure tumour size after doxorubicin with and without the candidate.

• Model: Doxorubicin cardiotoxicity in C57BL/6 or BALB/c mice. Use the dosing schedule from a published protocol.
• Groups (6 mice each): vehicle, doxorubicin alone, doxorubicin + dexrazoxane, and doxorubicin + each of the top 2 candidates from the egg screen, chosen before the mouse study starts. That is 30 mice. Reduce if your ethics committee requires.
• Readouts: Body weight, survival, serum troponin or CK-MB, heart weight to body weight ratio, ECG if available, and heart histology (vacuolisation and fibrosis).

Timeline (about 12 weeks)

What stops each score from being 10

• Novelty (8): the pieces exist. A second lab or a larger candidate set would raise it.
• Cost (9): doxorubicin, dexrazoxane and troponin kits are the main expenses.
• Feasibility (8): the dose-finding step can take a few attempts.
• Wow (9): it would reach 10 if one candidate protects the heart without protecting the tumour in both the egg and the mouse.

Backups

Caveats

• Ratings are my own, not published scores.
• Doses and regimens must come from published protocols and your ethics committee, not from me.
• Embryo rules differ by country, and the mouse work needs ethics approval.

1. Retinoblastoma (7.25)

• In ovo: Patient-derived retinoblastoma and choroidal melanoma xenografts have been established on the avian CAM (Kant et al., Indian J Ophthalmol 2023). Test drug candidates, including repurposed or natural agents, on the grafts.
• Mouse: A subcutaneous cell-line xenograft (for example Y79) in immunodeficient mice. Orthotopic eye models are technically difficult.
• Why it stands out: A childhood eye cancer with a personalised-medicine angle gives the highest wow factor.
• Risks: Patient tissue needs consent and ethics approval, and it is scarce. Start with cell lines. Immunodeficient mice cost more.

2. Hepatocellular carcinoma (7.0)

• In ovo: The CAM is an established HCC xenograft model (Li et al., JoVE 2015). A 2026 study used it to test a decorin-expressing herpes simplex virus vector against CAM-xenografted HCC (Frejborg et al.).
• Mouse: Chemically induced HCC (the DEN model) or a subcutaneous HCC cell xenograft. Measure tumour burden, liver enzymes, CD31 and histology.
• Why it ranks here: Safe and feasible, but the CAM-HCC pairing is already known, so your novelty depends on the compounds you test.

3. Ovarian cancer peritoneal spread (6.75)

• In ovo: CAM has been described as a cost-effective model for ovarian cancer metastasis (Chu et al., 2021). Measure invasion and spread, with human-DNA qPCR for quantification.
• Mouse: Intraperitoneal ovarian cancer cells (for example ID8 in C57BL/6 mice). Measure ascites volume, tumour nodules and survival.

4. Infantile hemangioma (6.75)

• In ovo: I found a chick CAM vascular tumour model induced by FGF-2 and inhibited by cidofovir (Liekens et al., 2001), but no direct CAM hemangioma model. You would need to adapt and validate it.
• Mouse: Hemangioma-stem-cell implants in immunodeficient mice, which are costly and technical.
• Why it stands out: Propranolol is the standard treatment, so you can use it as the positive control and test cheap alternatives.

5. Influenza antiviral (6.5)

• In ovo: The embryonated egg is an established in vivo antiviral model (Härtl et al., 2004), and ginger and garlic extracts have been tested against H9N2 in chick embryos (Rasool et al., 2017).
• Mouse: A mouse-adapted influenza strain (for example PR8, a lab strain). Measure weight loss, survival and lung viral titre.
• Risks: You need a virology lab and biosafety approval. Avian influenza strains are not suitable for most labs.

6. Implant-associated S. aureus infection (6.5)

• In ovo: Antibiotic-loaded bone cement has been tested on the CAM for biocompatibility (Ferreira et al., 2017). Add biofilm and infection readouts.
• Mouse: Subcutaneous or bone implant colonised with S. aureus. Measure bacterial counts on the implant and surrounding tissue.
• Risks: Needs microbiology facilities, and the surgery adds complexity.

7. Glioblastoma (6.25)

• In ovo: U87 glioblastoma xenografts on the CAM have been compared with rat brain xenografts (Strojnik et al., 2010), and there is a 2022 review of the model.
• Mouse: A subcutaneous U87 xenograft is the cheapest, but an orthotopic brain model is more relevant and needs stereotactic equipment.
• Why it ranks last: Higher cost, and the CAM-glioma model is already well established.

My picks from this list

• Best wow factor: Retinoblastoma, if you can get tissue or use cell lines.
• Most practical: Hepatocellular carcinoma.
• Strongest repurposing story: Infantile hemangioma, using propranolol as the benchmark.

1. The disease in brief

• Most children present with a white pupil (leukocoria), in 61% of cases, with strabismus in 22%. - Textbook of Family Medicine 9e (Retinoblastoma section)
• Most children are identified by 3 years of age, and family or friends usually notice the white pupil first. - Tintinalli's Emergency Medicine (Retinoblastoma section)
• The cause is loss of the RB1 tumour suppressor gene on chromosome 13q14.2. - Sabiston Textbook of Surgery (Hereditary retinoblastoma)
• The Rb protein controls the G1 to S transition by regulating E2F transcription factors. - Basic Medical Biochemistry, 6e (The Retinoblastoma gene)
• About 60% of cases are sporadic, with both RB1 alleles inactivated by chance in a single retinal cell. - Thompson & Thompson Genetics and Genomics in Medicine, 9e (Retinoblastoma)

2. The idea in one paragraph

3. What is already known (novelty check)

• Kant et al. 2023, Indian Journal of Ophthalmology grafted freshly cut patient retinoblastoma and choroidal melanoma onto the CAM at embryonic day 10 and incubated for 7 days. They saw angiogenesis around the graft and invasion into the CAM mesoderm. The tumours were synaptophysin and Ki-67 positive. The authors suggested the model could support personalised drug screening.
• Fialho et al. 2020 reported successful growth of fresh retinoblastoma cells on the CAM.
• Mouse models exist: intravitreal Y79 cells in nude mice (Tschulakow et al. 2016), a published orthotopic model for preclinical testing, and subcutaneous Y79 xenografts (Burr et al. 2011).

4. Design

• Start with a retinoblastoma cell line (for example Y79) rather than patient tissue. This avoids consent and supply problems.
• Confirm cell line identity before you start.
• Pilot the grafting. Follow the published schedule: window the egg, graft at embryonic day 10, harvest about 7 days later (Kant et al.). Record the graft take rate.
• Typical cell numbers per egg are in the order of a million cells in a gel plug or ring. Take the exact number from the cited papers and your pilot.

• Groups: vehicle, a standard retinoblastoma drug as the positive control (carboplatin is a reasonable choice because it is a standard retinoblastoma chemotherapy agent), and 4 to 5 candidates at two doses. That is about 12 groups with 8 eggs each, so about 100 eggs, and order about 125 for attrition.
• Candidates: plant compounds or repurposed drugs, whichever you can obtain.
• Treatment: topical or intratumoral dosing on 2 to 3 days after grafting.
• Readouts:
  • Tumour size from photographs in the egg-lab imaging box.
  • Vessels converging on the graft.
  • Ki-67 index, synaptophysin and invasion depth on histology.
  • Human DNA in the egg (Alu qPCR) for tumour burden, if you have a qPCR machine.
  • Embryo survival as a toxicity check.

• Model: Subcutaneous Y79 xenograft in immunodeficient (nude) mice. It is the cheapest and simplest. An intravitreal model is closer to the real disease but needs microinjection skill and imaging.
• Groups (6 mice each): vehicle, carboplatin, and the top two candidates from the egg screen, chosen and recorded before the mouse study begins. That is 24 mice.
• Readouts: Tumour volume over time by calipers, body weight, final tumour weight, Ki-67, synaptophysin and CD31.
• Prediction test: Did the egg ranking match the mouse ranking? With only a few candidates, report it as exploratory.

5. Timeline (about 13 weeks)

6. Why it is exciting

• It is a childhood eye cancer, and the story of an egg-based test saving mice is strong.
• A patient-derived version would be a personalised-medicine avatar, as Kant et al. suggest.
• The CAM result arrives in about a week, while mouse results take weeks.

7. Risks and limits

• Take rate: Y79 grows as suspension clusters, so grafts may not take reliably. The pilot will tell you.
• Mouse cost: Nude mice need special housing and are the biggest expense.
• Patient tissue: Scarce, and needs ethics approval and consent.
• Species gap: CAM tumours are in an avian host, and the egg lacks a mature immune system, so immune effects are not tested.
• Cytotoxic drugs: Handle carboplatin with proper protective equipment and waste disposal.
• Ethics: Embryo rules differ by country. The mouse work needs animal ethics approval.

8. Next step

Verified prices

• Y79 cell line (ATCC HTB-18): ATCC lists $577 for nonprofit customers, and Fisher Scientific lists $706.83.
• Fertilised eggs: A King's College London article says 12 fertilised eggs cost £45, with zero maintenance costs, a 97% saving compared with standard mouse xenografts. That is about £3.75 per egg. Local farm or hatchery eggs are often much cheaper.
• Nude mice: I could not retrieve a price. Vendors quote on request, and prices vary by country, so get a quote.

Budget by item

Where the money goes

• The mouse study is about half of the total, mainly the nude mice and their housing.
• Histology and antibodies are the second largest cost.
• Eggs are cheap. The egg screen is the lowest-cost part of the project.

Why the egg-first design saves money

Ways to reduce the cost

1. Get the cell line from a collaborator or a local cell repository instead of buying a new vial. This can save $500 to $700.
2. Use a subcutaneous xenograft, not an intravitreal one. It needs no microinjection equipment or eye imaging.
3. Use a core facility for histology and antibodies, which is often cheaper than buying antibodies for one project.
4. Skip the qPCR and measure tumour size and histology only.
5. Cut the egg groups to 8 and use 8 eggs per group.
6. Use 5 mice per group (20 mice) if your power calculation and ethics committee allow.
7. Build the imaging box from cardboard or 3D-print it.

Compared with the other topics

Retinoblastoma project cost (my estimate, in US dollars)

• Y79 cell line: ATCC lists $577 for nonprofit customers, and Fisher Scientific lists $706.83.
• Fertilised eggs: A King's College London article says 12 eggs cost £45, a 97% saving compared with standard mouse xenografts. That is about £3.75 per egg, and local hatchery eggs are often cheaper.
• Nude mice: I could not retrieve a price. Ask your supplier for a quote.

Where the money goes

• The mouse study is about half of the total.
• Histology and antibodies come second.
• The egg screen is the cheapest part.

Why the egg-first design saves money

Ways to lower the cost

1. Get the cell line from a collaborator or local repository, which can save $500 to $700.
2. Use a subcutaneous xenograft rather than an intravitreal one.
3. Use a core facility for histology and antibodies.
4. Skip the qPCR and measure tumour size and histology only.
5. Reduce the egg screen to 8 groups of 8 eggs.
6. Use 5 mice per group if your ethics committee and power calculation allow.
7. Build the imaging box from cardboard.

1. Anthracycline cardiomyopathy (8.5)

• In ovo: Phone-video heart rate in chick embryos treated with doxorubicin with or without cardioprotectants. Add a tumour graft on the CAM of parallel eggs to check the protectant does not shield the tumour.
• Mouse: Standard doxorubicin cardiotoxicity in common mice (30 animals). Measure troponin or CK-MB, heart weight and histology.
• Why it is feasible: No special mice or pathogens, and the egg part needs only a phone.

2. Diabetic embryopathy (8.0)

• In ovo: Hyperglycaemia in the chicken embryo is an established model. It has been used to study hyperglycaemia's effects on embryonic development (2025), congenital limb defects and a gestational diabetes eye malformation model involving Pax6 and oxidative stress. Your screen tests antioxidant or plant protectants: malformation rate, survival, and oxidative stress markers.
• Mouse: STZ-induced diabetic mice treated with the top candidate. Measure blood glucose, MDA, SOD and GSH, and kidney or liver histology. A pregnant-mouse arm is possible but adds cost and time.
• Why it works: The disease relevance is clear, STZ is a cheap and widely used model, and the egg part is quick.
• Novelty: The model exists, so novelty comes from your protectant and from linking embryo protection to the maternal diabetic markers.

3. Psoriasis (7.75)

• In ovo: CAM vessel and irritation screen of herbal anti-psoriatic candidates.
• Mouse: Imiquimod-induced psoriasis-like skin in BALB/c or C57BL/6 mice over about a week. Score erythema, scaling and skin thickness, plus histology.
• Caveat: I found no established CAM psoriasis model, so the egg reflects only the angiogenic and irritant parts of the disease.

4. Cancer with a dose-sparing twist (7.75)

• In ovo: Murine tumours, including Ehrlich carcinoma, have been grown on the CAM (Uchida et al., 1987). An Ehrlich tumour study with CAM work also exists (Pereira et al., 2013). Test drug and herbal combinations on CAM tumours at several doses.
• Mouse: Ehrlich solid tumour in Swiss albino mice, which are cheap and need no immunodeficient animals. Compare full-dose chemotherapy, half-dose chemotherapy plus adjunct, and controls.
• Twist: The question is whether an adjunct allows a lower chemotherapy dose with the same effect, which is a clinically meaningful goal.
• Novelty: Moderate, since the models are established.

5. Drug-induced liver injury (7.75)

• In ovo: Treat embryos with paracetamol with or without extracts. Score survival and liver injury markers (glutathione, MDA, histology).
• Mouse: Paracetamol hepatotoxicity in mice, which is cheap and takes days. Measure ALT, AST, glutathione and liver histology.
• Caveat: I found paracetamol liver injury modelled in Xenopus embryos but not in chick embryos, so the egg model would need a pilot to confirm it works. That makes it a bit riskier, but it is also where the novelty is.

6. Glucocorticoid-induced bone loss (7.25)

• In ovo: Whole-mount bone and cartilage staining of chick embryos is an established method (Yamazaki et al., 2011). Treat with a glucocorticoid with or without candidates and quantify bone staining from photographs. I did not find an existing osteoporosis screen of this kind, so this is a new application and needs validation.
• Mouse: Dexamethasone-induced bone loss. Measure serum calcium and alkaline phosphatase, femur ash weight and calcium, and histology. Micro-CT is optional.
• Caveat: Embryonic bone formation is not the same as adult bone loss, so state that limit plainly.

My picks

1. Safest and strongest: Idea 1 (doxorubicin cardiomyopathy).
2. Best story for a cheap model: Idea 2 (diabetic embryopathy).
3. Most straightforward mouse work: Idea 4 (Ehrlich tumour) or idea 3 (psoriasis).

1. Cancer: egg-lipid nanocarrier with self-fluorescent curcumin (8.25)

• Hook: Curcumin fluoresces, so you can watch the carrier travel through CAM blood vessels without adding a dye. The egg supplies the carrier material and the test system.
• What exists: Egg yolk lipids have been used to build a tumour-targeted nanovector (Tang et al., 2020), and chicken egg yolk nanovesicles have been characterised (Saadeldin et al., 2023). Your novelty is the live self-fluorescence tracking on the CAM and the CAM-to-mouse comparison, which I did not find.
• In ovo: Compare free curcumin with the loaded carrier for fluorescence in vessels over time and CAM tumour size.
• Mouse: Ehrlich solid tumour in Swiss albino mice. Compare free drug, carrier, and controls on tumour volume and histology.
• Needs: A blue LED or UV torch, a phone with a yellow filter, and a way to make lipid vesicles (thin-film or ethanol injection).

2. Cancer: kitchen photodynamic therapy (8.0)

• Hook: Use live CAM video to find the light dose that closes tumour vessels, then take that dose into the mouse. The vessel-shutdown video is very striking.
• What exists: The CAM is a known model for photodynamic therapy (Gottfried et al., 1992; quail CAM method, 2022), and curcumin photodynamic therapy on the CAM was published in 2019. So novelty is moderate. Your twist is a standardised light-dose finder and CAM-to-mouse dose transfer.
• Mouse: Ehrlich solid tumour in Swiss albino mice with photosensitiser and LED light, versus dark and light-only controls.
• Safety: Keep light exposures within a measured dose, and protect your eyes.

3. Epilepsy: embryo-movement AI screen (8.0)

• Hook: Neuroactive drugs change embryo spontaneous movement, and this was studied decades ago (Sedlacek, 1979). A phone and a motion-tracking script turn it into an objective screen.
• In ovo: Film embryo movement before and after convulsant and anticonvulsant exposure. Use reference drugs to prove the assay can tell the classes apart before you test your candidates.
• Mouse: PTZ seizure model. Record latency to seizure, seizure duration and mortality.
• Caveat: Embryo movement is not a seizure. State clearly that the egg is a neuroactivity filter, and validate the ranking against the mouse results.

4. Sepsis: embryo heart-rate early marker (7.75)

• Hook: Endotoxin toxicity in chick embryos was shown decades ago, with antibiotics such as polymyxins neutralising it (Rifkind and Palmer, 1966). The idea of heart rate as an early readout is my own suggestion and I have not verified it, so it needs a pilot.
• In ovo: LPS challenge with candidate treatments. Record survival and heart-rate change.
• Mouse: LPS endotoxaemia in mice. Measure survival, TNF-alpha and IL-6 by ELISA, and organ histology.

5. Cataract: phone slit-lamp transparency scoring (7.75)

• Hook: Embryonic chick lenses have been used as cataract models (Li et al., 2021; drug-induced cataract in cultured chick lens, 1973). The new part is cheap, objective phone-based opacity scoring plus a screen of antioxidants.
• Mouse: The classical selenite cataract model is done in rat pups (review). I did not confirm a mouse-pup protocol, so check this before choosing this topic if mice are required.

6. Snakebite: embryo-lethality screen for plant antivenoms (7.0)

• What exists: A hen's egg method has been proposed as a humane alternative to venom LD50 testing (non-neurotoxic venoms), a plant root extract neutralised viper venom in chick embryos, and egg yolk antibodies neutralised cobra and krait venom. Parts of this are not new.
• Hook: Neglected tropical disease, big public health impact, and a clear 3Rs story.
• Why it is last: Venom is hard to obtain and regulated, and the work needs trained handling. Feasibility is 5.

My picks

• Most creative with good feasibility: Idea 1 (egg-lipid nanocarrier).
• Best visual: Idea 2 (vascular-shutdown video).
• Newest disease area: Idea 3 (epilepsy).

1. Chemotherapy protection platform (8.5)

• Hook: A protectant is only useful if it does not also shield the tumour. The egg can test organ protection (heart rate by phone video, or kidney histology) and tumour response together.
• What exists: Cisplatin nephrotoxicity has been assessed in chick embryos (Náprstková et al., 2003). The combined protection-plus-tumour design is my own suggestion and I did not find it published, though my search was not exhaustive.
• In ovo: Embryos get the drug with or without candidates. Readouts: heart rate and survival (doxorubicin) or kidney weight and histology (cisplatin). Graft tumour cells on the CAM of a parallel set.
• Rat: Cisplatin acute kidney injury in SD rats (serum creatinine, BUN, kidney histology) or doxorubicin cardiotoxicity (troponin or CK-MB, heart histology). You can pick one organ for a shorter project.
• Optional: An Ehrlich tumour arm in Swiss mice to show the drug still works.

2. Hyperuricaemia and gout (8.25)

• Hook: Birds make uric acid as their main nitrogen waste, so the embryo is a ready-made uric acid factory. Embryonic xanthine oxidase activity and its inhibition have been studied (Duggan et al., 1968; Butros and Khalidi, 1966), but those are old studies. A cheap modern plant-extract screen with a standard uric acid assay is the new angle. I did not find one.
• In ovo: Treat embryos with allopurinol (reference) and plant extracts. Measure uric acid in embryo blood or allantoic fluid, plus survival and malformations.
• Mouse: Swiss albino mice given potassium oxonate (which blocks uricase, since mice, unlike birds, break uric acid down) with or without hypoxanthine. Measure serum uric acid, liver xanthine oxidase activity, creatinine and kidney histology. A rat paw urate-crystal model is an option for the gout flare.
• Caveat: Embryo purine metabolism differs from adult mammals, so use allopurinol as a benchmark and a pilot to confirm the assay works.

3. Myocardial infarction (8.0)

• What exists: Embryonic heart models of anoxia-reoxygenation have been used (Bruchez et al., 2008), and hypoxia causes dilated cardiomyopathy in chick embryos (Tintu et al., 2009). Day-4 embryos can keep their heart rate during hypoxia (Nechaeva et al., 2020), so stage and hypoxia severity need a pilot.
• Hook: Phone-video heart-rate recovery after hypoxia as a quick screen for candidate extracts.
• Rat: Isoproterenol-induced myocardial injury in SD rats (a few days). Measure troponin or CK-MB, ECG if available, and heart histology or TTC staining.

4. Breast cancer prevention: the egg reads the rat's cancer risk (7.25)

• Hook: Mammary tissue at different stages elicits different angiogenic responses on the CAM (Strum, 1983, mouse tissue), and DMBA-induced lesions are angiogenic (Heffelfinger et al., 2000). So take small biopsies from DMBA-treated SD rats at set times, test their angiogenic potency on the CAM, and see whether a chemopreventive diet or extract lowers it before tumours appear.
• Rat: SD female rats given DMBA, with chemopreventive treatment. Follow tumour incidence and latency.
• Why it ranks lower: The model takes months and DMBA is a hazardous carcinogen needing proper handling. Feasibility is 6.

5. Neonatal brain injury (stretch, 7.0)

• What exists: A hypoxic-ischaemic injury model in the embryonic chicken cerebral pallium has been described (Olivares-Hernández et al., 2022).
• Rat: The pup model needs carotid ligation surgery and a controlled hypoxia chamber. I rate feasibility at 5 because of the surgical skill and equipment.

Earlier ideas that already fit your animals

• Egg-lipid curcumin nanocarrier with live CAM tracking, validated in Ehrlich or Dalton's lymphoma tumours in Swiss mice (8.25). Dalton's lymphoma-induced angiogenesis has been studied before (Kumari et al., 2017).
• Kitchen photodynamic therapy with vascular-shutdown video, validated in Ehrlich solid tumours (8.0).
• Embryo-movement AI anticonvulsant screen, validated in the PTZ seizure model (8.0).
• Embryo sepsis screen, validated in LPS endotoxaemia in Swiss mice (7.75).
• Cataract screen on embryonic lens, validated in selenite-induced cataract. This is classically done in rat pups, so SD pups should work (7.75), but confirm the strain in the protocol you follow.
• Diabetic embryopathy, validated in STZ-diabetic rats (8.0).
• Doxorubicin cardioprotection in SD rats (8.5).

My picks

1. Most creative and still practical: Idea 1 (chemotherapy protection platform).
2. Newest disease area: Idea 2 (hyperuricaemia and gout).
3. Quickest rat work: Idea 3 (myocardial infarction).

Why this pairing works

Ranked ideas

The ideas in brief

• In ovo: Photograph CAMs with a smartphone on a cheap macro lens. Train a free Python U-Net or use AngioTool to count vessels automatically.
• In vivo: Run a mouse Matrigel plug or rat wound, scored by haemoglobin content and CD31 staining.
• Novelty: A cheap, open-source tool that shows CAM image scores predict in vivo angiogenesis. A correlation figure (r, Bland-Altman) is the headline result.

• Scaffold: Calcine waste eggshell into hydroxyapatite and blend it with chitosan or gelatin. Add a plant osteogenic extract or a trace-ion dopant such as Sr or Zn for extra novelty.
• In ovo: Check biocompatibility and angiogenesis on the CAM.
• In vivo: Test in a rat calvarial defect, with micro-CT if you can get it.
• Why it stands out: The circular-economy story is easy to remember.

• In ovo: Inject at day 0 to 3 and score mortality, malformations, heart rate and CAM vascularity.
• In vivo: Run a dose-matched rat or zebrafish study for liver, kidney and oxidative stress markers.
• Why it stands out: Public interest is high. Adding a rescue arm (for example an antioxidant) makes it mechanistic.

• In ovo: Create a high-glucose environment that impairs angiogenesis, then test pro-angiogenic candidates (honey, curcumin, PRP, a peptide).
• In vivo: Take the best candidate into an STZ-diabetic rat excision wound.
• Why it stands out: A high-glucose CAM screen is a quick way to test diabetic-wound candidates.

• In ovo: Infect the CAM with S. aureus, Pseudomonas or Candida, then treat with essential oil, phage or an antimicrobial peptide. Score bacterial load and lesion size.
• In vivo: Use a mouse or rat infected wound.
• Why it stands out: Antimicrobial resistance is a strong theme.

• Design: Screen 8 to 10 extracts by MTT assay, then CAM, then one tumour model (Ehrlich ascites or a xenograft).
• Novelty: Report how well the in ovo ranking predicted the in vivo ranking (Spearman correlation). This is the safest and cheapest option. It becomes more novel if you choose an under-studied regional plant.

• Dressing: Electrospun or cast film containing anthocyanin from red cabbage, which changes colour with wound pH, plus an active such as curcumin.
• In ovo: HET-CAM irritation test and angiogenesis.
• In vivo: Rat wound, tracking pH against healing rate.
• Why it stands out: The colour change is visual and photographs well.

• Material: Make fluorescent carbon dots from banana peel, curry leaf or similar by hydrothermal synthesis.
• In ovo: Image them live through a window in the shell, tracking vessel uptake and tumour accumulation.
• In vivo: Run a mouse biodistribution study.
• Why it stands out: Live glowing vessels make a strong figure. It needs a fluorescence imager or UV box with a camera, which is the main cost and risk.

• Setup: A cheap LED array plus a photosensitiser such as curcumin or methylene blue.
• In ovo: Treat CAM tumours or vessels.
• In vivo: Use a mouse subcutaneous tumour.
• Why it stands out: It frames the work as affordable cancer therapy for low-resource settings.

• In ovo: Measure embryonic heart rate and rhythm by video with doxorubicin, then add a protective natural compound.
• In vivo: Rat doxorubicin model with ECG and cardiac enzymes.

• In ovo: Graft patient-derived or cell-line tumour fragments onto the CAM and test drug combinations.
• In vivo: Compare with a mouse xenograft.
• Why it stands out: The personalised-medicine angle is very strong, but it needs tissue consent and ethics approval, so it is harder. Cell lines are a safer start.

My picks

• Highest wow with low risk: Idea 2, the eggshell scaffold.
• Best for publication: Idea 1, because validation work is rare and tool-based papers get cited.
• Safest for a student timeline: Idea 6, or idea 4 for something slightly more novel.

Practical notes

• Eggs: Use fertilised eggs, incubated at 37.5°C and about 60% humidity. Use about 8 to 10 eggs per group. Plan for 20 to 30% attrition.
• Ethics: Rules for embryos differ by country and institution, so confirm with your ethics committee. Mouse and rat work needs animal ethics approval. Zebrafish larvae are often treated more leniently.
• Minimal controls: Vehicle, positive control (for example bevacizumab or sunitinib for anti-angiogenesis) and untreated.
• Statistics: One-way ANOVA with a post hoc test, plus the in ovo vs in vivo correlation.

Details

• Extract features from every CAM image: vessel density, branching points, lacunarity and fractal dimension. Combine them with dose and compound descriptors.
• Train a simple model (random forest or regression) to predict the in vivo result, such as wound closure percentage or tumour volume. Validate it on held-out compounds.
• "We predicted mouse results from an egg" is a strong headline, and it costs only your time. The risk is sample size, so you need at least 10 to 15 compounds or doses.

• In ovo: Inject probiotics, prebiotics, plant extracts or nanominerals into the egg on about day 17 to 18. This is a real poultry technique. Record hatchability and hatch weight.
• In vivo: Raise the hatched chicks for 3 to 6 weeks. Measure growth, gut villus height, microbiota, antibody titre after vaccination and heat-stress markers.
• Why it stands out: There is no rodent colony and no cross-species gap. A continuous egg-to-bird story is easy to explain, and the poultry industry may fund it.

• In ovo: Puncture a standard-size CAM vessel and apply chitosan, alum, a plant extract or gelatin sponge. Time the bleeding to stop and quantify blood loss from images.
• In vivo: Rat tail transection or liver-wound bleeding.
• Why it stands out: The cheap, standardised in ovo bleeding assay is the novelty. The CAM puncture model is much less used than the angiogenesis assay.

• In ovo: Test classical herbo-mineral preparations for embryo toxicity, malformations and CAM effects.
• In vivo: Sub-acute rat study of the safest and the most toxic preparations (liver and kidney enzymes, histology, tissue metal levels).
• Why it stands out: It addresses a real gap in the safety evidence for traditional medicines. Add metal quantification (ICP-MS or AAS) if your lab can reach it.

• Setup: A low-cost red/near-infrared LED array with a power meter.
• In ovo: Test several wavelengths and doses on CAM angiogenesis and embryo viability.
• In vivo: Rat or diabetic-rat wound, with the optimal dose from the CAM.
• Why it stands out: It finds the best dose cheaply before committing animals.

• Patch: Make dissolvable microneedles (PVA, gelatin or sugar) in a PDMS mould cast from a cheap commercial needle array or a 3D-printed master.
• In ovo: Test needle penetration and dye diffusion on the CAM. Embryo skin or CAM also works.
• In vivo: Test insulin or a drug in rats.
• Risk: The moulds and fabrication take trial and error.

• In ovo: Expose eggs to a phone or Wi-Fi router during incubation. Record hatchability, malformations, heart rate and oxidative stress.
• In vivo: Rats exposed for several weeks, with a protective antioxidant arm if you want a mechanistic angle.
• Caution: Past results in this field are mixed and often poorly controlled. Use sham exposure, a measured field strength and blinded scoring.

• In ovo: Measure venom-induced haemorrhage in embryos and test plant extracts as inhibitors.
• In vivo: Mouse neutralisation assays.
• Why it stands out: Embryo screening could cut the number of mice used in venom neutralisation work.
• Risk: Venom is hard to obtain and regulated, so I rated it low on feasibility.

Combinations for the most impact

• Eggshell hydroxyapatite scaffold (from the first list) + AI vessel quantification + digital twin (A). The story is egg waste to a bone graft, assessed by an open-source tool and predicted by a model.
• Hyperglycaemic CAM + LED photobiomodulation (E) + digital twin (A). This is an affordable diabetic-wound platform, with the dose chosen in ovo.
• Haemostat (C) + smart dressing (from the first list). A first-aid kit concept.

Next step

Details

• In ovo: Film the embryo through a window or by candling with a phone. Extract heart rate and motion with free signal-processing code (Eulerian video magnification or simple intensity tracking). Use it as a non-invasive, repeatable toxicity or cardiotoxicity marker over days.
• In vivo: Compare with rat ECG after the same drug (doxorubicin, a QT-prolonging drug or an energy drink).
• Why it stands out: You replace expensive instruments with a phone and open code. The key step is validating it against a reference method.

• In ovo: Expose embryos or the CAM to UV-B or UV-C through a film or cream. Score survival, vessel damage and oxidative markers.
• In vivo: Mouse or rat UV skin model (erythema, thickness, MDA and histology).
• Extra: Measure SPF in vitro by spectrophotometer. A UV lamp with a measured dose is the main equipment need.

• In ovo: Grow tumour spheroids, 3D-printed or hydrogel scaffolds, or stem cell constructs on the CAM for several days so chick vessels grow in.
• In vivo: Implant the prevascularised construct in a mouse or rat and test whether it survives better than an unvascularised one.
• Why it stands out: It is a creative use of the egg as a cheap incubator for tissue engineering. The main risk is that the vessels are avian, so check for host integration and rejection.

• In ovo: Graft endometrial cells or tissue and test anti-angiogenic or anti-adhesion agents. Ovarian cancer metastasis on the CAM is well described, but verify the endometriosis protocols in the literature before committing.
• In vivo: Mouse endometriosis model.
• Why it stands out: Women's health is under-funded and reviewers like the topic. Use cell lines first, because human tissue needs ethics approval and consent.

• In ovo: Induce hypoxia with a low-oxygen setup or chemically (for example CoCl2). Test protective agents such as rhodiola, ginkgo or antioxidants on survival, vascular density and HIF-1α.
• In vivo: Rat hypobaric chamber if available, or a chemical hypoxia model (sodium nitrite).

• In ovo: Treat with valproate or another teratogen, then add folic acid, choline or a plant antioxidant. Score neural tube defects, limb and heart malformations, and size.
• In vivo: Pregnant rat with the best rescue dose.
• Why it stands out: The rescue design gives you a mechanistic story. Pregnant-animal work has stricter ethics and a longer timeline.

• In ovo: A CAM inflammation model is cheap, so screen many extracts.
• In vivo: Carrageenan rat paw oedema.
• Note: This is the easiest and safest option, but the least exciting. It works well as a base for idea 1 from the previous list, the AI quantification.

• In ovo: Treat with lead, cadmium or arsenic and test moringa, chlorella or coriander extract.
• In vivo: Rat model with blood and tissue metal levels and liver and kidney markers.

Ways to raise the wow factor in any project

• Time-lapse imaging. A phone on a clamp or a 3D-printed holder can film CAM vessel growth over 24 to 72 hours.
• Open data and code. Posting images and scripts (for example on Zenodo or GitHub) raises citation chances.
• Preregistration of your design and endpoints, which reviewers value.
• A single visual summary. A clear egg-to-animal workflow figure and one striking before and after image.
• A frugal comparison. Show cost per experiment against a mouse-only design.
• Negative controls and blinding. Score images blinded, which adds credibility at no cost.

Next step

1. Open-source Frugal Egg Lab kit (9.0)

• Concept: A complete, cheap, reproducible kit for CAM experiments. It has a 3D-printed or cardboard imaging box with a phone clamp and LED ring, a standard window-cutting and sealing protocol, and free software that counts vessels.
• In ovo: Run an anti-angiogenesis and a pro-angiogenesis screen with known controls (bevacizumab or sunitinib, and VEGF or a pro-angiogenic extract).
• In vivo: Validate against one mouse Matrigel plug or rat wound experiment, scored by haemoglobin and CD31.
• Why it scores high: Labs with no imaging equipment can use it, and it is publishable as a methods paper.
• What holds it back: Image consistency across labs. To get to 10, test the kit in a second lab and report inter-lab agreement.

2. Public benchmark plus digital twin (8.75)

• Concept: Build the first small open benchmark of CAM results for 15 to 20 compounds whose in vivo potency is already published (known anti-angiogenic and pro-angiogenic drugs). Then test whether a simple model predicts the published in vivo ranking.
• In ovo: Dose-response of each compound on the CAM.
• In vivo: Mostly literature data, plus one small confirmatory experiment with a new compound the model never saw.
• Why it scores high: Almost no animals are needed, and a dataset others can reuse is a durable contribution.
• What holds it back: Published in vivo data are inconsistent across species and doses. To reach 10, preregister the prediction and test it on the unseen compound.

3. Egg-to-bird feeding of waste-derived bioactives (8.75)

• Concept: Inject waste-derived bioactives (eggshell calcium nanoparticles, fruit-peel polyphenols or similar) into the egg at about day 17 to 18, then raise the chicks.
• In ovo: Hatchability, hatch weight and embryo toxicity.
• In vivo: The same birds for 3 to 6 weeks: growth, gut villus height, antibody response to vaccination and heat-stress markers.
• Why it scores high: There is no rodent colony, the cost is the lowest of all seven, and the poultry industry may support it.
• What holds it back: Novelty is moderate because in ovo feeding is established. To reach 10, pair a new waste-derived compound with a measurable health outcome such as vaccine response.

4. Everyday exposome in an egg (8.5)

• Concept: Test substances people use daily: tea-bag microplastic leachate, instant-noodle cup leachate, vape liquid, energy drinks and hair-dye ingredients. Compare them on one standard embryo scoring system.
• In ovo: Mortality, malformations, heart rate (from video) and CAM vascularity.
• In vivo: Rat or zebrafish for the one or two worst offenders, with liver, kidney and oxidative stress endpoints.
• Why it scores high: Everyone can relate to it, and press and social media will pick it up.
• What holds it back: Real-world dose relevance. Dose-match to realistic human exposure to avoid overclaiming.

5. Eggshell-to-bone with AI scoring (8.5)

• Concept: Calcine eggshell waste into hydroxyapatite and blend it with chitosan or gelatin, with or without a plant osteogenic extract.
• In ovo: Biocompatibility and angiogenesis on the CAM, scored with the idea 1 pipeline.
• In vivo: Rat calvarial defect with micro-CT or histology.
• Why it scores high: The "egg to bone" story gives the strongest visual and narrative of all seven.
• What holds it back: Eggshell hydroxyapatite is known, and imaging adds cost. To reach 10, add a clear twist, such as a trace-ion dopant, and use the CAM score to select the formulation.

6. Kitchen-to-clinic haemostat (8.5)

• Concept: Compare cheap haemostats: chitosan from shrimp-shell waste, alum, turmeric, gelatin sponge and honey.
• In ovo: Puncture a standard CAM vessel and time the bleeding to stop. Quantify blood loss from images.
• In vivo: Rat tail transection or liver-wound bleeding.
• Why it scores high: Very low cost, a clear practical use in low-resource first aid, and an easy outcome to measure.
• What holds it back: Standardising the puncture. To reach 10, use a fixed needle gauge, a template and blinded scoring.

7. Diabetic-wound platform, stacked (8.0)

• Concept: Combine a hyperglycaemic CAM, LED photobiomodulation dose-finding and a pH-indicator dressing.
• In ovo: Screen dressing candidates and LED doses under high glucose.
• In vivo: Diabetic rat wound (STZ), tracking healing rate and wound pH.
• Why it scores high: It is a cohesive story with several outputs.
• What holds it back: It needs more skills (dressing fabrication, LED dosimetry and the diabetic model) and a larger scope, so it is better as a PhD plan than a short project.

My recommendation

• AngioTool is a free NIH tool for vessel length, density, branching and lacunarity (Zudaire et al., PLoS One 2011).
• A commercial IKOSA CAM app does deep-learning vessel segmentation (Kuri et al., Cells 2022).
• An AI-based CAM angiogenesis method was published in 2022 (Faihs et al.).

What it is

1. A standard imaging box. It is cardboard or 3D-printed, with a matte white interior, a diffuse LED ring, a fixed phone holder at a fixed height, and a ruler and colour card in every frame.
2. A standard CAM protocol. It covers eggs, windowing, how the test substance is applied, timing and scoring rules.
3. Free analysis software. A Python or ImageJ pipeline gives vessel area %, total length, branch points and lacunarity as a CSV file.
4. Validation. You compare it with manual counting, AngioTool and one in vivo experiment.

Workflow

1. Incubate fertilised eggs at about 37.5°C and about 60% humidity.
2. On about day 3, cut a small window, or use an ex-ovo shell-less culture.
3. On about day 7 to 8, place a filter-paper disc, silicone ring or gelatin sponge with the test substance on the CAM.
4. Photograph at about day 10 to 12 in the box. Fix the settings and take a photo of the colour card each time.
5. Count only eggs that survive. Report attrition.

• Vehicle control.
• Pro-angiogenic control (bFGF or VEGF).
• Anti-angiogenic control (sunitinib or suramin).

• Start with the green channel, where blood vessels show best contrast.
• Baseline: a Frangi vessel filter with thresholding.
• Optional: train a small U-Net on 30 to 50 hand-annotated images.
• Outputs: vessel area %, total length, branch points and lacunarity.

• Mouse Matrigel plug: Measure haemoglobin (Drabkin) and CD31 staining. This is the standard comparison for angiogenesis.
• Rat excision wound: Use it if you are testing a healing compound. Measure wound closure and granulation tissue.

Example design (my suggestion)

• In ovo: 3 test compounds x 3 doses + 3 controls = 12 groups, with 8 to 10 eggs per group. That is about 100 eggs, so plan for about 130 with attrition.
• In vivo: 4 groups (vehicle, positive control, two doses of the best candidate) with 6 mice per group, or about 24 animals. Run a formal power calculation once you have pilot variance.

Validation steps

Timeline (about 12 weeks)

Cost

Outputs

• A methods paper or short report.
• An open GitHub repository with the box design, code and a step-by-step protocol.
• A shared image dataset with annotations.
• A ready-made teaching tool for other labs.

Risks

• Lighting and glare are the biggest threat to reproducibility. Use polarising film and always include the colour card.
• Egg variability. Source eggs from one supplier and use fixed incubation conditions.
• Annotation effort. Manual labelling takes time. Start with the Frangi baseline and only add the U-Net if it improves agreement.
• Ethics. Embryo rules vary by country, and mouse or rat work needs animal ethics approval.
• Novelty pushback. Cite AngioTool and IKOSA, and state clearly that your contribution is the open, validated, low-cost kit.

Next step

The mouse model I recommend: the Matrigel plug assay

How to do it

1. Pick the test set from the egg results. Choose compounds that cover the CAM range: one strong effect, one moderate effect and one with no effect. Add a vehicle control and a known positive control. This is what lets you test whether the egg ranking predicts the mouse ranking.
2. Preregister the prediction. Write down the expected order before the mouse study begins.
3. Prepare the plugs. Mix cold liquid Matrigel with heparin and an angiogenic factor (bFGF or VEGF), plus the test compound or vehicle. For a pro-angiogenic compound, add it alone and compare with the vehicle plug.
4. Inject the mixture subcutaneously (for example the flank) with a cold syringe. It gels at body temperature.
5. Harvest at about day 7 to 14 (use the timing in your chosen protocol).
6. Read out with:
   • Photographs in the egg-lab box (colour and redness).
   • Haemoglobin content (Drabkin method), the cheap quantitative readout.
   • CD31 immunostaining if you have the antibody, as the histology confirmation.

Group sizes

Options if you cannot use Matrigel

• Matrigel cost. It can be the largest single expense, so get a local quote early. Dilute it only as the protocol allows.
• Directed in vivo angiogenesis assay (DIVAA). The silicone-tube kits are also available, but the plug is cheaper and simpler.
• Mouse excision wound. Use this if your compounds are wound-healing agents. It needs no Matrigel and is cheaper. Measure closure, granulation tissue and hydroxyproline.
• Ehrlich ascites or solid tumour in Swiss albino mice. Use this if your compounds are anticancer. Measure tumour volume, survival and vessel density. It is widely used in low-resource labs, but it measures anticancer effect more than angiogenesis.

Analysis plan

• Compare groups with one-way ANOVA and a post hoc test, or Kruskal-Wallis if the data are skewed.
• Test whether the CAM ranking matches the mouse ranking with a Spearman correlation or simple rank agreement.
• With only 3 to 4 compounds, a correlation coefficient is weak. Add doses (for example 2 doses of each compound), report effect sizes with confidence intervals, and describe the correlation as exploratory.
• Report the full results, including cases where the egg and the mouse disagree.

Mouse study timeline (about 4 weeks)

Ethics and approval

• Get approval from your institutional animal ethics committee before you start (in India this is the IAEC under CPCSEA). Rules differ by country, so check yours.
• State the 3Rs argument: the egg screen reduces the number of compounds tested in mice, so fewer animals are used.
• Set humane endpoints and keep the procedure short. This model involves a small subcutaneous injection and a short experiment.

Updated project summary

1. Kombucha-cellulose "living" wound dressing (8.5)

• Concept: The cellulose pellicle that grows on kombucha (the SCOBY) is a cheap bacterial cellulose sheet. Purify it, load it with a plant bioactive such as curcumin, and optionally add a pH-indicator dye (anthocyanin) so it changes colour as the wound pH shifts.
• In ovo: CAM biocompatibility, irritation and angiogenesis.
• Mouse: Full-thickness excision wound (splinted if available). Measure closure rate, histology and wound pH.
• Why it works: The material costs almost nothing and the story is memorable ("tea fungus to skin patch").

2. Frugal drug repurposing for anti-angiogenesis (8.25)

• Concept: Screen 15 to 20 cheap, off-patent drugs (examples of candidates: itraconazole, mebendazole, doxycycline, propranolol, metformin, disulfiram) on the CAM. Pick the best for mouse testing.
• In ovo: Dose-response, scored with the Frugal Egg Lab kit.
• Mouse: Matrigel plug, or Ehrlich or B16 tumour if you want an anticancer angle.
• Why it works: Drugs with known safety profiles are easier to take forward, and it answers a practical question for low-resource settings. Check that the mouse dose is a sensible human-equivalent dose.

3. Sentinel egg for food adulterants and cosmetics (8.25)

• Concept: Use embryos as a cheap toxicity screen for things consumers worry about. Examples: metanil yellow in spices, formalin in fish, carbide-ripened fruit extracts, skin-lightening creams, kajal. Adapt the list to your country.
• In ovo: Embryo survival, malformations, heart rate by phone video (see the earlier video-readout idea) and CAM vascularity.
• Mouse: Short sub-acute oral or dermal study of the worst one or two (liver and kidney enzymes, histology).
• Why it works: Direct public relevance and strong press potential. Use realistic exposure doses, to avoid alarming conclusions.

4. Whole-egg dressing (8.25)

• Concept: Use all the egg: an eggshell membrane sheet, an egg-white (ovalbumin) hydrogel, and calcium from the shell. Compare them alone and combined.
• In ovo: CAM angiogenesis and biocompatibility.
• Mouse: Excision wound.
• Why it works: The material and the test model both come from the egg. Eggshell membrane is a known biomaterial, so your novelty is the combination and the CAM-to-mouse validation.

5. Phage cocktail from local sewage (7.75)

• Concept: Isolate bacteriophages from sewage or river water that kill wound pathogens (for example Pseudomonas aeruginosa, S. aureus) and formulate them in a gel.
• In ovo: CAM infection model with treatment. Score lesion size, bacterial load and embryo survival.
• Mouse: Infected wound model.
• Why it works: Antimicrobial resistance is a strong theme and the raw material is free.
• Risks: You need a microbiology lab and appropriate biosafety approval for the pathogens. Phage isolation takes skill.

6. Smokeless tobacco and areca nut toxicity (7.75)

• Concept: Compare extracts of regional products (gutka, khaini, areca nut, betel quid) at realistic doses.
• In ovo: Embryo toxicity, CAM vascular effects and oxidative stress markers.
• Mouse: Oral mucosa or cheek exposure over several weeks, with histology and a protective antioxidant arm if you want a mechanism.
• Why it works: Strong local public health relevance. Long exposure studies take time, so plan the timeline.

7. Regional honey passport (7.75)

• Concept: Collect honeys from different regions or floral sources and compare them: phenolics, antioxidants and angiogenic activity.
• In ovo: CAM dose-response.
• Mouse: Excision wound with the best and worst honey.
• Why it works: Cheap, easy and locally meaningful. Make it more novel by profiling the chemistry (total phenolics, a few marker compounds) and linking it to the wound result.

8. Edible plant nanovesicles (7.5)

• Concept: Isolate nanosized vesicles from ginger, grape, wheatgrass or lemon juice. Test them on the CAM for angiogenesis and tissue effects.
• Mouse: Wound healing or a colitis model.
• Why it works: This is a fashionable research area with a strong wow factor.
• Risks: Isolation by ultracentrifugation needs equipment. Precipitation-based isolation is cheaper but less pure, so plan to characterise your vesicles (size by DLS or electron microscopy).

My picks

• Best overall: Idea 1, the kombucha dressing. It is cheap, visual and simple to run, and the mouse wound model is well established.
• Best for public interest: Idea 3, the sentinel egg.
• Best fit with the Frugal Egg Lab kit: Idea 2, drug repurposing, because the egg screen and mouse plug are the project.

• Novelty: how new the combination is. I cap it at 9.
• Cost: 10 means cheapest.
• Feasibility: can one student or lab finish it in 3 to 4 months.
• Wow: visual impact and story appeal.

Topics and ratings

1. Kombucha cellulose dressing (9.25)

• In ovo: CAM biocompatibility, irritation and angiogenesis, to choose the best curcumin and dye loading.
• Mouse: Excision wound with groups for untreated, plain cellulose, commercial dressing and the loaded dressing. Measure closure rate, histology and wound pH.
• Scores:
  • Novelty 8: bacterial cellulose dressings exist, but the pH indicator plus CAM-guided optimisation is a fresh combination.
  • Cost 10: the material is a by-product of kombucha.
  • Feasibility 9: simple chemistry and a standard wound model.
  • Wow 10: the dressing changes colour as the wound changes, and it photographs well.
• To raise novelty to 9: show that the colour change predicts healing outcome.

2. Sentinel egg for food adulterants (9.0)

• In ovo: Embryo survival, malformations and heart rate from phone video, for 5 to 8 locally relevant substances at realistic doses.
• Mouse: Sub-acute oral or dermal exposure to the worst one or two (liver and kidney enzymes, histology).
• Scores:
  • Novelty 8: embryo toxicity testing is known, but the phone-based readout and the local adulterant focus are new.
  • Cost 9: eggs and a phone.
  • Feasibility 9: short experiments.
  • Wow 10: it speaks directly to what people eat and use.
• To raise novelty to 9: validate against a reference substance panel with known toxicity.

3. Open egg-to-mouse angiogenesis platform (9.0)

• In ovo: CAM screen of 10 to 15 compounds with the Frugal Egg Lab kit.
• Mouse: Matrigel plug with vehicle, positive control and three compounds spanning the CAM range. Measure haemoglobin and CD31.
• Scores:
  • Novelty 8: tools like AngioTool and IKOSA exist, so the novelty is the open, validated, low-cost kit.
  • Cost 10: almost no equipment.
  • Feasibility 9: well-established assays.
  • Wow 9: strong methods and data story.
• To raise novelty to 9: test the kit in a second lab.

4. CAM-based digital twin for wound dressings (9.0)

• In ovo: CAM screen of several dressing formulations, with image features extracted by the kit.
• Mouse: Excision wound for a subset of formulations. Test whether the model, trained on the CAM data, predicted the healing ranking.
• Scores:
  • Novelty 9: predicting mouse outcome from egg images is uncommon.
  • Cost 9: the cost is mostly the mouse study.
  • Feasibility 8: it needs enough formulations (at least 8 to 10) for the model.
  • Wow 10: "we predicted mouse healing from an egg."
• To raise feasibility to 9: use simple models, preregister the prediction and keep the formulation set small but varied.

5. Egg-waste hydroxyapatite for bone (8.75)

• In ovo: CAM angiogenesis and integration of scaffold variants.
• Mouse: Critical-size calvarial defect, assessed by micro-CT if available and histology.
• Scores:
  • Novelty 8: eggshell hydroxyapatite is known, so use a twist such as a trace-ion dopant.
  • Cost 9: waste material, but imaging adds cost.
  • Feasibility 8: surgery and imaging are more demanding.
  • Wow 10: "egg to bone."

What I recommend

Novelty correction

• Kombucha-derived cellulose wound dressings have been published (Dey et al., 2024).
• Anthocyanin-loaded bacterial cellulose pH indicators exist (Zubova et al., 2024; Melnyk et al., 2025).

Title

Aims

1. Make and characterise kombucha cellulose dressings with different curcumin and anthocyanin loadings.
2. Screen them on the chick CAM for irritation and angiogenesis.
3. Test selected formulations in a mouse excisional wound model.
4. Test whether the CAM data predicted the mouse ranking, and whether dressing colour tracks wound pH and healing.

1. Dressing preparation (weeks 1-3)

1. Grow the pellicle. Use a kombucha culture in sweetened tea at room temperature, static, for about 1 to 2 weeks.
2. Purify. Wash, treat with dilute NaOH with heat to remove cells, then rinse to neutral pH. Sterilise by autoclave or another method that does not damage the dye.
3. Load the actives. Soak in curcumin (dissolved in a small amount of solvent) and/or red-cabbage anthocyanin extract.
4. Characterise. Record weight, thickness and water uptake. Measure release into PBS over 24 hours by UV-Vis. Photograph the colour at known buffer pH values (for example 4 to 9) to build a calibration curve.

2. In ovo screen (weeks 3-6)

• Groups: 8 formulations + untreated + a commercial dressing + a pro-angiogenic control (bFGF or VEGF disc) = 11 groups.
• Eggs: 10 per group is 110, so order about 140 with attrition.
• Method: Window the egg early. Apply a disc-sized piece of dressing on the CAM at about day 7 to 8. Photograph at about 48 to 72 hours in the egg-lab imaging box.
• Irritation: Run a separate HET-CAM test on day 9 to 10 eggs (score haemorrhage, vessel lysis and coagulation over a few minutes).
• Readouts: Vessel area %, total length, branch points, lacunarity, avascular zone, irritation score and embryo survival.
• Analysis: Blinded scoring, ANOVA or Kruskal-Wallis, and report all groups, including failures.

3. Mouse experiment (weeks 6-10)

• Wound photographs at days 0, 3, 5, 7, 10 and 14, with a ruler and colour card in the egg-lab box.
• Dressing colour (hue or RGB) at each time point.
• Wound surface pH if you have a flat pH probe or strips.
• Histology at day 14: H&E for re-epithelialisation and granulation thickness, CD31 for vessel density.

4. Digital-twin analysis (Topic 4)

• Inputs: CAM features per formulation (vessel area %, branching, lacunarity, irritation score) plus the curcumin and anthocyanin levels.
• Target: Mean mouse wound closure per formulation.
• Models: Keep them simple (ridge regression or random forest) with two or three features chosen in advance. Use leave-one-formulation-out cross-validation.
• Benchmarks: Compare against predicting the average, and against a model using only curcumin dose. If the CAM features do not add accuracy over curcumin dose, say so.
• Reporting: Spearman rank correlation and mean absolute error.
• Honest limit: With only 4 to 6 mouse formulations, treat this as exploratory. Preregister the predictions before the mouse study unblinds.
• Colour analysis: Correlate dressing hue with surface pH and with closure.

Timeline

Cost

Risks and fixes

• Contamination of the pellicle: Culture in clean vessels, sterilise before use.
• Batch variability: Grow all dressing pieces from the same batches and record their thickness.
• Colour drift: Fix lighting and use the colour card in every photograph.
• Few formulations for the model: Add more formulations to the CAM screen even if only some go to the mouse.
• Dressing adherence in mice: Use the standard secondary cover and check dressing position daily.

Expected outputs

• A dressing recipe and its characterisation.
• A CAM dataset with images, features and scores.
• A mouse healing dataset.
• A prediction analysis and open code.

Draft abstract (to fill with your results)

1. Sentinel egg for food adulterants (9.0)

• In ovo: Test 5 to 8 locally relevant substances at realistic doses. Record survival, malformations and heart rate from phone video.
• Mouse: Sub-acute oral or dermal dosing of the worst one or two. Measure liver and kidney enzymes and histology.

2. Embryo heart-rate video screen for cardioprotection (8.0)

• In ovo: Treat embryos with doxorubicin, with and without plant compounds, and film the heart. Chick embryos have already been used to detect drug cardiotoxicity (for example trastuzumab in chick embryos), so the novelty is the cheap video readout and the protectant screen.
• Mouse: Doxorubicin cardiotoxicity with the top candidate. Measure ECG or echo if available, troponin or CK-MB, and cardiac histology.

3. Egg-grown oncolytic Newcastle disease virus (7.75)

• In ovo: The egg is the cheap way to grow the virus. Then test its effect on tumour cells grafted on the CAM.
• Mouse: Syngeneic or xenograft tumour with intratumoral virus.
• Why it stands out: The egg does two jobs, and oncolytic virus therapy is an exciting area. My PubMed search only returned old virology papers on NDV growth in the CAM, so I could not confirm how novel the CAM tumour angle is.
• Risks: NDV is a poultry pathogen. Use a mild vaccine-type strain, get biosafety approval and contain it away from any birds. This is why feasibility is only 6.

4. Drug-combination synergy on CAM tumours (7.75)

• In ovo: Graft tumour cells and test 2-drug combinations at several doses. Compare tumour size or weight.
• Mouse: Ehrlich tumour or a syngeneic tumour with the best combination against each drug alone.
• Why it stands out: Combination screening is expensive in mice and cheap in eggs. Calculate synergy with a standard method (Bliss or Loewe).

5. Antithrombotic natural product screen (7.5)

• In ovo: Induce a small vessel injury on the CAM and measure clot formation or flow. My search did not find an established CAM thrombosis model, so you would need to develop and validate it. That gives novelty but adds risk.
• Mouse: Tail bleeding time (to check for excess bleeding) plus collagen/epinephrine-induced pulmonary thromboembolism survival.

6. Herbal UV photoprotection (7.5)

• In ovo: Expose embryos or CAM to a measured UV dose through a thin film of formulation. Score survival, vessel damage and oxidative markers.
• Mouse: UVB skin exposure. Measure erythema, skin thickness and MDA, plus histology.
• Needs: A UV lamp and a UV meter.

7. High-glucose embryo to diabetic mouse (7.0)

• In ovo: Inject glucose to create a high-glucose environment and test protective compounds on vessel and embryo outcomes.
• Mouse: STZ diabetic mice. Measure retinal or renal vascular leakage (for example Evans blue) and kidney histology. This model needs weeks, so plan the timeline.

8. Anti-metastasis CAM screen (6.75)

• In ovo: CAM metastasis methods are established (Maacha and Saule) and a 2026 paper reports digital PCR for quantification, but digital PCR is expensive. Regular qPCR for human DNA is the cheaper route.
• Mouse: B16F10 tail-vein injection and lung nodule counts.
• Why it ranks last: Novelty is low because the method is established.

My picks

• Highest overall: Idea 1, the sentinel egg.
• Best balance of safety and feasibility: Idea 2, the cardioprotection screen.
• Biggest wow: Idea 3, the oncolytic virus, if you can handle the biosafety requirements.

1. Anthracycline cardiomyopathy (8.0)

• In ovo: Treat embryos with doxorubicin with or without candidate compounds. Film the heart with a phone and extract heart rate and rhythm. Chick embryos have been used for cardiotoxicity testing before (trastuzumab example), so your novelty is the cheap video readout and a protectant screen.
• Mouse: Doxorubicin cardiotoxicity regimen from a published protocol, with the best candidate. Measure ECG or echo if available, troponin or CK-MB, and cardiac histology.
• Why it ranks first: Short experiments and clear endpoints.

2. Psoriasis (7.75)

• In ovo: CAM vessel and irritation scoring to rank candidates.
• Mouse: Topical imiquimod on shaved back skin of BALB/c or C57BL/6 mice for about a week. Score erythema, scaling and thickness (a PASI-like score) and examine epidermal thickening on histology. The model is short and cheap (example study).
• Caveat: My search found no established CAM psoriasis model. The CAM only captures the angiogenic and irritant components of the disease, so state that limit clearly.

3. Diabetic retinopathy (7.25)

• In ovo: There is precedent for hyperglycaemia-induced angiogenesis defects in the CAM (Larger et al., Diabetes 2004) and for a CAM/human-vitreous platform in diabetic retinopathy (Rezzola et al., 2020).
• Mouse: STZ-induced diabetes in C57BL/6 mice, with retinal vascular leakage (Evans blue) or acellular capillary counts on retinal flat-mounts.
• Risk: The diabetic phase takes several weeks, so the timeline is longer.

4. Oral squamous cell carcinoma (7.25)

• In ovo: CAM tumour growth and vessel assays. CAM has even been tried in the clinical workflow with patient tissue (Kauffmann et al., 2018). If you use patient samples, you need ethics approval and consent.
• Mouse: Cell-line xenograft in immunodeficient mice (shorter but costly) or the 4NQO carcinogenesis model (cheaper but takes months).
• Why it stands out: Strong relevance where tobacco and areca nut use is common.

5. Endometriosis (7.0)

• In ovo: Human endometrial fragments form endometriosis-like lesions on the CAM (Maas et al., 2001), and angiostatic agents and MMP inhibitors reduced lesions there (Nap et al., 2005). The model is established, so novelty comes from what you test.
• Mouse: Syngeneic uterine-tissue transplant into the peritoneum. Measure lesion size, vascularity and histology.
• Note: Human tissue needs consent and approval.

6. Candidiasis (7.0)

• In ovo: Infect the CAM or embryo and score survival and fungal burden. I did not verify the specific published CAM Candida protocol, so check the literature first.
• Mouse: Oral or vaginal candidiasis, with fungal CFU counts and histology.
• Needs: A microbiology lab and appropriate biosafety approval.

7. Triple-negative breast cancer (6.75)

• In ovo: MDA-MB-231 tumours on the CAM, treated with drug pairs (checkerboard design).
• Mouse: 4T1 syngeneic tumour in BALB/c mice. Track tumour growth and lung metastasis.
• Why it ranks lower: CAM breast tumour work is well established, so novelty depends on your drug combinations.

8. Rheumatoid arthritis (6.5)

• In ovo: CAM has been used to compare anti-angiogenic effects in RA and osteoarthritis synovium (Ribatti et al., 2000).
• Mouse: Collagen-induced arthritis in DBA/1 mice (more costly) or a cheaper adjuvant-induced model. Score paw swelling, clinical arthritis score and joint histology.

9. Retinopathy of prematurity (6.5)

• In ovo: CAM angiogenesis screen of candidates.
• Mouse: Neonatal pups exposed to high oxygen, then room air, with retinal flat-mount to measure neovascular tufts and avascular area.
• Risk: It needs an oxygen chamber and neonatal mouse handling.

My picks

• Easiest and fastest: Idea 1 (anthracycline cardiomyopathy) or idea 2 (psoriasis).
• Best story: Idea 4 (oral cancer), if you can handle the longer mouse timeline.
• Established models, so low risk: Ideas 5 and 7.

Best topic: doxorubicin cardiomyopathy

Novelty correction

• Smartphone measurement of chick embryo heart rate (Phuphanin et al., 2019).
• In ovo video of the beating heart used to study antiarrhythmic drug effects (Kockova et al., 2013).
• A chick embryo toxicity test compared with rodents for 20 drugs (Nishigori et al., 1992).

The key idea (the wow factor)

• Measure the embryo's heart rate and rhythm (cardiac effect).
• Graft tumour cells on the CAM of the same eggs and measure tumour size (anti-cancer effect).

Design

• Groups: vehicle, doxorubicin alone, doxorubicin + dexrazoxane (the approved cardioprotectant, as the positive control), and doxorubicin + 4 to 6 candidates (for example curcumin, quercetin, resveratrol or a local plant extract). Add candidate-alone groups for toxicity. Use about 10 eggs per group.
• Dose-finding pilot: Find a doxorubicin dose that lowers heart rate or survival without killing most embryos.
• Readouts: Heart rate by phone video, beat-to-beat variability, survival, malformation score and, if possible, an oxidative stress marker (MDA) in embryo tissue.
• Tumour check: Graft tumour cells on the CAM in a parallel set of eggs. Measure tumour size after doxorubicin with and without the candidate.

• Model: Doxorubicin cardiotoxicity in C57BL/6 or BALB/c mice. Use the dosing schedule from a published protocol.
• Groups (6 mice each): vehicle, doxorubicin alone, doxorubicin + dexrazoxane, and doxorubicin + each of the top 2 candidates from the egg screen, chosen before the mouse study starts. That is 30 mice. Reduce if your ethics committee requires.
• Readouts: Body weight, survival, serum troponin or CK-MB, heart weight to body weight ratio, ECG if available, and heart histology (vacuolisation and fibrosis).

Timeline (about 12 weeks)

What stops each score from being 10

• Novelty (8): the pieces exist. A second lab or a larger candidate set would raise it.
• Cost (9): doxorubicin, dexrazoxane and troponin kits are the main expenses.
• Feasibility (8): the dose-finding step can take a few attempts.
• Wow (9): it would reach 10 if one candidate protects the heart without protecting the tumour in both the egg and the mouse.

Backups

Caveats

• Ratings are my own, not published scores.
• Doses and regimens must come from published protocols and your ethics committee, not from me.
• Embryo rules differ by country, and the mouse work needs ethics approval.

1. Retinoblastoma (7.25)

• In ovo: Patient-derived retinoblastoma and choroidal melanoma xenografts have been established on the avian CAM (Kant et al., Indian J Ophthalmol 2023). Test drug candidates, including repurposed or natural agents, on the grafts.
• Mouse: A subcutaneous cell-line xenograft (for example Y79) in immunodeficient mice. Orthotopic eye models are technically difficult.
• Why it stands out: A childhood eye cancer with a personalised-medicine angle gives the highest wow factor.
• Risks: Patient tissue needs consent and ethics approval, and it is scarce. Start with cell lines. Immunodeficient mice cost more.

2. Hepatocellular carcinoma (7.0)

• In ovo: The CAM is an established HCC xenograft model (Li et al., JoVE 2015). A 2026 study used it to test a decorin-expressing herpes simplex virus vector against CAM-xenografted HCC (Frejborg et al.).
• Mouse: Chemically induced HCC (the DEN model) or a subcutaneous HCC cell xenograft. Measure tumour burden, liver enzymes, CD31 and histology.
• Why it ranks here: Safe and feasible, but the CAM-HCC pairing is already known, so your novelty depends on the compounds you test.

3. Ovarian cancer peritoneal spread (6.75)

• In ovo: CAM has been described as a cost-effective model for ovarian cancer metastasis (Chu et al., 2021). Measure invasion and spread, with human-DNA qPCR for quantification.
• Mouse: Intraperitoneal ovarian cancer cells (for example ID8 in C57BL/6 mice). Measure ascites volume, tumour nodules and survival.

4. Infantile hemangioma (6.75)

• In ovo: I found a chick CAM vascular tumour model induced by FGF-2 and inhibited by cidofovir (Liekens et al., 2001), but no direct CAM hemangioma model. You would need to adapt and validate it.
• Mouse: Hemangioma-stem-cell implants in immunodeficient mice, which are costly and technical.
• Why it stands out: Propranolol is the standard treatment, so you can use it as the positive control and test cheap alternatives.

5. Influenza antiviral (6.5)

• In ovo: The embryonated egg is an established in vivo antiviral model (Härtl et al., 2004), and ginger and garlic extracts have been tested against H9N2 in chick embryos (Rasool et al., 2017).
• Mouse: A mouse-adapted influenza strain (for example PR8, a lab strain). Measure weight loss, survival and lung viral titre.
• Risks: You need a virology lab and biosafety approval. Avian influenza strains are not suitable for most labs.

6. Implant-associated S. aureus infection (6.5)

• In ovo: Antibiotic-loaded bone cement has been tested on the CAM for biocompatibility (Ferreira et al., 2017). Add biofilm and infection readouts.
• Mouse: Subcutaneous or bone implant colonised with S. aureus. Measure bacterial counts on the implant and surrounding tissue.
• Risks: Needs microbiology facilities, and the surgery adds complexity.

7. Glioblastoma (6.25)

• In ovo: U87 glioblastoma xenografts on the CAM have been compared with rat brain xenografts (Strojnik et al., 2010), and there is a 2022 review of the model.
• Mouse: A subcutaneous U87 xenograft is the cheapest, but an orthotopic brain model is more relevant and needs stereotactic equipment.
• Why it ranks last: Higher cost, and the CAM-glioma model is already well established.

My picks from this list

• Best wow factor: Retinoblastoma, if you can get tissue or use cell lines.
• Most practical: Hepatocellular carcinoma.
• Strongest repurposing story: Infantile hemangioma, using propranolol as the benchmark.

1. The disease in brief

• Most children present with a white pupil (leukocoria), in 61% of cases, with strabismus in 22%. - Textbook of Family Medicine 9e (Retinoblastoma section)
• Most children are identified by 3 years of age, and family or friends usually notice the white pupil first. - Tintinalli's Emergency Medicine (Retinoblastoma section)
• The cause is loss of the RB1 tumour suppressor gene on chromosome 13q14.2. - Sabiston Textbook of Surgery (Hereditary retinoblastoma)
• The Rb protein controls the G1 to S transition by regulating E2F transcription factors. - Basic Medical Biochemistry, 6e (The Retinoblastoma gene)
• About 60% of cases are sporadic, with both RB1 alleles inactivated by chance in a single retinal cell. - Thompson & Thompson Genetics and Genomics in Medicine, 9e (Retinoblastoma)

2. The idea in one paragraph

3. What is already known (novelty check)

• Kant et al. 2023, Indian Journal of Ophthalmology grafted freshly cut patient retinoblastoma and choroidal melanoma onto the CAM at embryonic day 10 and incubated for 7 days. They saw angiogenesis around the graft and invasion into the CAM mesoderm. The tumours were synaptophysin and Ki-67 positive. The authors suggested the model could support personalised drug screening.
• Fialho et al. 2020 reported successful growth of fresh retinoblastoma cells on the CAM.
• Mouse models exist: intravitreal Y79 cells in nude mice (Tschulakow et al. 2016), a published orthotopic model for preclinical testing, and subcutaneous Y79 xenografts (Burr et al. 2011).

4. Design

• Start with a retinoblastoma cell line (for example Y79) rather than patient tissue. This avoids consent and supply problems.
• Confirm cell line identity before you start.
• Pilot the grafting. Follow the published schedule: window the egg, graft at embryonic day 10, harvest about 7 days later (Kant et al.). Record the graft take rate.
• Typical cell numbers per egg are in the order of a million cells in a gel plug or ring. Take the exact number from the cited papers and your pilot.

• Groups: vehicle, a standard retinoblastoma drug as the positive control (carboplatin is a reasonable choice because it is a standard retinoblastoma chemotherapy agent), and 4 to 5 candidates at two doses. That is about 12 groups with 8 eggs each, so about 100 eggs, and order about 125 for attrition.
• Candidates: plant compounds or repurposed drugs, whichever you can obtain.
• Treatment: topical or intratumoral dosing on 2 to 3 days after grafting.
• Readouts:
  • Tumour size from photographs in the egg-lab imaging box.
  • Vessels converging on the graft.
  • Ki-67 index, synaptophysin and invasion depth on histology.
  • Human DNA in the egg (Alu qPCR) for tumour burden, if you have a qPCR machine.
  • Embryo survival as a toxicity check.

• Model: Subcutaneous Y79 xenograft in immunodeficient (nude) mice. It is the cheapest and simplest. An intravitreal model is closer to the real disease but needs microinjection skill and imaging.
• Groups (6 mice each): vehicle, carboplatin, and the top two candidates from the egg screen, chosen and recorded before the mouse study begins. That is 24 mice.
• Readouts: Tumour volume over time by calipers, body weight, final tumour weight, Ki-67, synaptophysin and CD31.
• Prediction test: Did the egg ranking match the mouse ranking? With only a few candidates, report it as exploratory.

5. Timeline (about 13 weeks)

6. Why it is exciting

• It is a childhood eye cancer, and the story of an egg-based test saving mice is strong.
• A patient-derived version would be a personalised-medicine avatar, as Kant et al. suggest.
• The CAM result arrives in about a week, while mouse results take weeks.

7. Risks and limits

• Take rate: Y79 grows as suspension clusters, so grafts may not take reliably. The pilot will tell you.
• Mouse cost: Nude mice need special housing and are the biggest expense.
• Patient tissue: Scarce, and needs ethics approval and consent.
• Species gap: CAM tumours are in an avian host, and the egg lacks a mature immune system, so immune effects are not tested.
• Cytotoxic drugs: Handle carboplatin with proper protective equipment and waste disposal.
• Ethics: Embryo rules differ by country. The mouse work needs animal ethics approval.

8. Next step

Verified prices

• Y79 cell line (ATCC HTB-18): ATCC lists $577 for nonprofit customers, and Fisher Scientific lists $706.83.
• Fertilised eggs: A King's College London article says 12 fertilised eggs cost £45, with zero maintenance costs, a 97% saving compared with standard mouse xenografts. That is about £3.75 per egg. Local farm or hatchery eggs are often much cheaper.
• Nude mice: I could not retrieve a price. Vendors quote on request, and prices vary by country, so get a quote.

Budget by item

Where the money goes

• The mouse study is about half of the total, mainly the nude mice and their housing.
• Histology and antibodies are the second largest cost.
• Eggs are cheap. The egg screen is the lowest-cost part of the project.

Why the egg-first design saves money

Ways to reduce the cost

1. Get the cell line from a collaborator or a local cell repository instead of buying a new vial. This can save $500 to $700.
2. Use a subcutaneous xenograft, not an intravitreal one. It needs no microinjection equipment or eye imaging.
3. Use a core facility for histology and antibodies, which is often cheaper than buying antibodies for one project.
4. Skip the qPCR and measure tumour size and histology only.
5. Cut the egg groups to 8 and use 8 eggs per group.
6. Use 5 mice per group (20 mice) if your power calculation and ethics committee allow.
7. Build the imaging box from cardboard or 3D-print it.

Compared with the other topics

Retinoblastoma project cost (my estimate, in US dollars)

• Y79 cell line: ATCC lists $577 for nonprofit customers, and Fisher Scientific lists $706.83.
• Fertilised eggs: A King's College London article says 12 eggs cost £45, a 97% saving compared with standard mouse xenografts. That is about £3.75 per egg, and local hatchery eggs are often cheaper.
• Nude mice: I could not retrieve a price. Ask your supplier for a quote.

Where the money goes

• The mouse study is about half of the total.
• Histology and antibodies come second.
• The egg screen is the cheapest part.

Why the egg-first design saves money

Ways to lower the cost

1. Get the cell line from a collaborator or local repository, which can save $500 to $700.
2. Use a subcutaneous xenograft rather than an intravitreal one.
3. Use a core facility for histology and antibodies.
4. Skip the qPCR and measure tumour size and histology only.
5. Reduce the egg screen to 8 groups of 8 eggs.
6. Use 5 mice per group if your ethics committee and power calculation allow.
7. Build the imaging box from cardboard.

1. Anthracycline cardiomyopathy (8.5)

• In ovo: Phone-video heart rate in chick embryos treated with doxorubicin with or without cardioprotectants. Add a tumour graft on the CAM of parallel eggs to check the protectant does not shield the tumour.
• Mouse: Standard doxorubicin cardiotoxicity in common mice (30 animals). Measure troponin or CK-MB, heart weight and histology.
• Why it is feasible: No special mice or pathogens, and the egg part needs only a phone.

2. Diabetic embryopathy (8.0)

• In ovo: Hyperglycaemia in the chicken embryo is an established model. It has been used to study hyperglycaemia's effects on embryonic development (2025), congenital limb defects and a gestational diabetes eye malformation model involving Pax6 and oxidative stress. Your screen tests antioxidant or plant protectants: malformation rate, survival, and oxidative stress markers.
• Mouse: STZ-induced diabetic mice treated with the top candidate. Measure blood glucose, MDA, SOD and GSH, and kidney or liver histology. A pregnant-mouse arm is possible but adds cost and time.
• Why it works: The disease relevance is clear, STZ is a cheap and widely used model, and the egg part is quick.
• Novelty: The model exists, so novelty comes from your protectant and from linking embryo protection to the maternal diabetic markers.

3. Psoriasis (7.75)

• In ovo: CAM vessel and irritation screen of herbal anti-psoriatic candidates.
• Mouse: Imiquimod-induced psoriasis-like skin in BALB/c or C57BL/6 mice over about a week. Score erythema, scaling and skin thickness, plus histology.
• Caveat: I found no established CAM psoriasis model, so the egg reflects only the angiogenic and irritant parts of the disease.

4. Cancer with a dose-sparing twist (7.75)

• In ovo: Murine tumours, including Ehrlich carcinoma, have been grown on the CAM (Uchida et al., 1987). An Ehrlich tumour study with CAM work also exists (Pereira et al., 2013). Test drug and herbal combinations on CAM tumours at several doses.
• Mouse: Ehrlich solid tumour in Swiss albino mice, which are cheap and need no immunodeficient animals. Compare full-dose chemotherapy, half-dose chemotherapy plus adjunct, and controls.
• Twist: The question is whether an adjunct allows a lower chemotherapy dose with the same effect, which is a clinically meaningful goal.
• Novelty: Moderate, since the models are established.

5. Drug-induced liver injury (7.75)

• In ovo: Treat embryos with paracetamol with or without extracts. Score survival and liver injury markers (glutathione, MDA, histology).
• Mouse: Paracetamol hepatotoxicity in mice, which is cheap and takes days. Measure ALT, AST, glutathione and liver histology.
• Caveat: I found paracetamol liver injury modelled in Xenopus embryos but not in chick embryos, so the egg model would need a pilot to confirm it works. That makes it a bit riskier, but it is also where the novelty is.

6. Glucocorticoid-induced bone loss (7.25)

• In ovo: Whole-mount bone and cartilage staining of chick embryos is an established method (Yamazaki et al., 2011). Treat with a glucocorticoid with or without candidates and quantify bone staining from photographs. I did not find an existing osteoporosis screen of this kind, so this is a new application and needs validation.
• Mouse: Dexamethasone-induced bone loss. Measure serum calcium and alkaline phosphatase, femur ash weight and calcium, and histology. Micro-CT is optional.
• Caveat: Embryonic bone formation is not the same as adult bone loss, so state that limit plainly.

My picks

1. Safest and strongest: Idea 1 (doxorubicin cardiomyopathy).
2. Best story for a cheap model: Idea 2 (diabetic embryopathy).
3. Most straightforward mouse work: Idea 4 (Ehrlich tumour) or idea 3 (psoriasis).

1. Cancer: egg-lipid nanocarrier with self-fluorescent curcumin (8.25)

• Hook: Curcumin fluoresces, so you can watch the carrier travel through CAM blood vessels without adding a dye. The egg supplies the carrier material and the test system.
• What exists: Egg yolk lipids have been used to build a tumour-targeted nanovector (Tang et al., 2020), and chicken egg yolk nanovesicles have been characterised (Saadeldin et al., 2023). Your novelty is the live self-fluorescence tracking on the CAM and the CAM-to-mouse comparison, which I did not find.
• In ovo: Compare free curcumin with the loaded carrier for fluorescence in vessels over time and CAM tumour size.
• Mouse: Ehrlich solid tumour in Swiss albino mice. Compare free drug, carrier, and controls on tumour volume and histology.
• Needs: A blue LED or UV torch, a phone with a yellow filter, and a way to make lipid vesicles (thin-film or ethanol injection).

2. Cancer: kitchen photodynamic therapy (8.0)

• Hook: Use live CAM video to find the light dose that closes tumour vessels, then take that dose into the mouse. The vessel-shutdown video is very striking.
• What exists: The CAM is a known model for photodynamic therapy (Gottfried et al., 1992; quail CAM method, 2022), and curcumin photodynamic therapy on the CAM was published in 2019. So novelty is moderate. Your twist is a standardised light-dose finder and CAM-to-mouse dose transfer.
• Mouse: Ehrlich solid tumour in Swiss albino mice with photosensitiser and LED light, versus dark and light-only controls.
• Safety: Keep light exposures within a measured dose, and protect your eyes.

3. Epilepsy: embryo-movement AI screen (8.0)

• Hook: Neuroactive drugs change embryo spontaneous movement, and this was studied decades ago (Sedlacek, 1979). A phone and a motion-tracking script turn it into an objective screen.
• In ovo: Film embryo movement before and after convulsant and anticonvulsant exposure. Use reference drugs to prove the assay can tell the classes apart before you test your candidates.
• Mouse: PTZ seizure model. Record latency to seizure, seizure duration and mortality.
• Caveat: Embryo movement is not a seizure. State clearly that the egg is a neuroactivity filter, and validate the ranking against the mouse results.

4. Sepsis: embryo heart-rate early marker (7.75)

• Hook: Endotoxin toxicity in chick embryos was shown decades ago, with antibiotics such as polymyxins neutralising it (Rifkind and Palmer, 1966). The idea of heart rate as an early readout is my own suggestion and I have not verified it, so it needs a pilot.
• In ovo: LPS challenge with candidate treatments. Record survival and heart-rate change.
• Mouse: LPS endotoxaemia in mice. Measure survival, TNF-alpha and IL-6 by ELISA, and organ histology.

5. Cataract: phone slit-lamp transparency scoring (7.75)

• Hook: Embryonic chick lenses have been used as cataract models (Li et al., 2021; drug-induced cataract in cultured chick lens, 1973). The new part is cheap, objective phone-based opacity scoring plus a screen of antioxidants.
• Mouse: The classical selenite cataract model is done in rat pups (review). I did not confirm a mouse-pup protocol, so check this before choosing this topic if mice are required.

6. Snakebite: embryo-lethality screen for plant antivenoms (7.0)

• What exists: A hen's egg method has been proposed as a humane alternative to venom LD50 testing (non-neurotoxic venoms), a plant root extract neutralised viper venom in chick embryos, and egg yolk antibodies neutralised cobra and krait venom. Parts of this are not new.
• Hook: Neglected tropical disease, big public health impact, and a clear 3Rs story.
• Why it is last: Venom is hard to obtain and regulated, and the work needs trained handling. Feasibility is 5.

My picks

• Most creative with good feasibility: Idea 1 (egg-lipid nanocarrier).
• Best visual: Idea 2 (vascular-shutdown video).
• Newest disease area: Idea 3 (epilepsy).

1. Chemotherapy protection platform (8.5)

• Hook: A protectant is only useful if it does not also shield the tumour. The egg can test organ protection (heart rate by phone video, or kidney histology) and tumour response together.
• What exists: Cisplatin nephrotoxicity has been assessed in chick embryos (Náprstková et al., 2003). The combined protection-plus-tumour design is my own suggestion and I did not find it published, though my search was not exhaustive.
• In ovo: Embryos get the drug with or without candidates. Readouts: heart rate and survival (doxorubicin) or kidney weight and histology (cisplatin). Graft tumour cells on the CAM of a parallel set.
• Rat: Cisplatin acute kidney injury in SD rats (serum creatinine, BUN, kidney histology) or doxorubicin cardiotoxicity (troponin or CK-MB, heart histology). You can pick one organ for a shorter project.
• Optional: An Ehrlich tumour arm in Swiss mice to show the drug still works.

2. Hyperuricaemia and gout (8.25)

• Hook: Birds make uric acid as their main nitrogen waste, so the embryo is a ready-made uric acid factory. Embryonic xanthine oxidase activity and its inhibition have been studied (Duggan et al., 1968; Butros and Khalidi, 1966), but those are old studies. A cheap modern plant-extract screen with a standard uric acid assay is the new angle. I did not find one.
• In ovo: Treat embryos with allopurinol (reference) and plant extracts. Measure uric acid in embryo blood or allantoic fluid, plus survival and malformations.
• Mouse: Swiss albino mice given potassium oxonate (which blocks uricase, since mice, unlike birds, break uric acid down) with or without hypoxanthine. Measure serum uric acid, liver xanthine oxidase activity, creatinine and kidney histology. A rat paw urate-crystal model is an option for the gout flare.
• Caveat: Embryo purine metabolism differs from adult mammals, so use allopurinol as a benchmark and a pilot to confirm the assay works.

3. Myocardial infarction (8.0)

• What exists: Embryonic heart models of anoxia-reoxygenation have been used (Bruchez et al., 2008), and hypoxia causes dilated cardiomyopathy in chick embryos (Tintu et al., 2009). Day-4 embryos can keep their heart rate during hypoxia (Nechaeva et al., 2020), so stage and hypoxia severity need a pilot.
• Hook: Phone-video heart-rate recovery after hypoxia as a quick screen for candidate extracts.
• Rat: Isoproterenol-induced myocardial injury in SD rats (a few days). Measure troponin or CK-MB, ECG if available, and heart histology or TTC staining.

4. Breast cancer prevention: the egg reads the rat's cancer risk (7.25)

• Hook: Mammary tissue at different stages elicits different angiogenic responses on the CAM (Strum, 1983, mouse tissue), and DMBA-induced lesions are angiogenic (Heffelfinger et al., 2000). So take small biopsies from DMBA-treated SD rats at set times, test their angiogenic potency on the CAM, and see whether a chemopreventive diet or extract lowers it before tumours appear.
• Rat: SD female rats given DMBA, with chemopreventive treatment. Follow tumour incidence and latency.
• Why it ranks lower: The model takes months and DMBA is a hazardous carcinogen needing proper handling. Feasibility is 6.

5. Neonatal brain injury (stretch, 7.0)

• What exists: A hypoxic-ischaemic injury model in the embryonic chicken cerebral pallium has been described (Olivares-Hernández et al., 2022).
• Rat: The pup model needs carotid ligation surgery and a controlled hypoxia chamber. I rate feasibility at 5 because of the surgical skill and equipment.

Earlier ideas that already fit your animals

• Egg-lipid curcumin nanocarrier with live CAM tracking, validated in Ehrlich or Dalton's lymphoma tumours in Swiss mice (8.25). Dalton's lymphoma-induced angiogenesis has been studied before (Kumari et al., 2017).
• Kitchen photodynamic therapy with vascular-shutdown video, validated in Ehrlich solid tumours (8.0).
• Embryo-movement AI anticonvulsant screen, validated in the PTZ seizure model (8.0).
• Embryo sepsis screen, validated in LPS endotoxaemia in Swiss mice (7.75).
• Cataract screen on embryonic lens, validated in selenite-induced cataract. This is classically done in rat pups, so SD pups should work (7.75), but confirm the strain in the protocol you follow.
• Diabetic embryopathy, validated in STZ-diabetic rats (8.0).
• Doxorubicin cardioprotection in SD rats (8.5).

My picks

1. Most creative and still practical: Idea 1 (chemotherapy protection platform).
2. Newest disease area: Idea 2 (hyperuricaemia and gout).
3. Quickest rat work: Idea 3 (myocardial infarction).