10 Fresh & Publishable CT/MRI Research Ideas

1. Dual-Energy CT (DECT) for Bone Marrow Edema vs. MRI - Diagnostic Accuracy Study

• Why it's fresh: DECT fat/edema maps are now validated in limited body parts. Extending this to new anatomical areas or specific populations (diabetic foot, pediatric) fills a direct gap (Radiology, 2025).
• Design: Prospective diagnostic accuracy study
• Feasibility: Mid-tier, doable with ~80-100 patients

2. Ultrashort Echo Time (UTE) MRI of the Lungs vs. Low-Dose CT

• Why it's fresh: UTE-MRI for lungs is an emerging area with growing but still limited original data. Comparison with LDCT in a specific disease subset (e.g., post-COVID fibrosis, pediatric) is novel.
• Design: Observational comparative study
• Feasibility: Requires MRI with UTE sequence availability

3. Photon-Counting CT vs. Conventional CT for Characterizing Incidental Adrenal Nodules

• Why it's fresh: PCD-CT launched commercially in 2023-2024; original clinical studies are very limited and high in demand right now.
• Design: Retrospective/prospective comparative study
• Feasibility: Needs access to PCD-CT scanner (major hospital)

4. Radiomics-Based CT Texture Analysis for Predicting Lymph Node Metastasis - Without Biopsy

• Why it's fresh: Radiomics is booming but most studies are single-center, small, and not validated externally. A well-designed study with histopathological ground truth is publishable in journals like European Radiology or Cancer Imaging.
• Design: Retrospective, with internal/external validation
• Feasibility: High - uses archived CT data + surgical histology reports

5. MRI-Based Body Composition Analysis (Sarcopenia/Fat Infiltration) as a Predictor of Surgical Outcomes

• Why it's fresh: Body composition from CT is established but MRI-based quantification is less explored and avoids radiation. A prospective correlative study adds genuine novelty.
• Design: Prospective correlational
• Feasibility: High - uses existing pre-op MRI scans

6. AI-Assisted Detection of Incidental Findings on Routine CT - Impact on Clinical Workflow

• Why it's fresh: AI tools for incidentaloma detection are entering clinical use but real-world audit data on missed findings and downstream management is scarce.
• Design: Prospective audit/observational
• Feasibility: Moderate - needs approved AI software deployed in your department

7. Synthetic MRI (SyMRI) vs. Conventional Multi-Sequence MRI for Brain Lesion Characterization

• Why it's fresh: Synthetic MRI reduces scan time significantly, but head-to-head accuracy data for specific pathologies in routine clinical populations is limited.
• Design: Prospective comparative
• Feasibility: Needs SyMRI software/sequence at your center

8. Low-Field Portable MRI (0.064T or 1T) vs. High-Field MRI in Post-ICU or Bedside Patients

• Why it's fresh: Portable low-field MRI (e.g., Hyperfine Swoop) is FDA-cleared but real-world head-to-head data in specific patient populations is still emerging rapidly.
• Design: Prospective comparative
• Feasibility: Needs portable MRI access

9. CT-Based Bone Mineral Density (Opportunistic Osteoporosis Screening) in Incidental Abdominal CTs

• Why it's fresh: Opportunistic screening using existing CT data is an active area with strong public health relevance. Publication in Osteoporosis International or AJR is very feasible.
• Design: Retrospective cross-sectional
• Feasibility: Very high - no new scans needed, just ROI measurement on archived data

10. MRI Diffusion Tensor Imaging (DTI) Tractography vs. CT for Pre-Surgical Mapping Near Eloquent Cortex

• Why it's fresh: DTI is used in neurosurgery but large-sample comparative studies with surgical outcome data remain limited. A multi-site or single-center prospective design is publishable.
• Design: Prospective observational
• Feasibility: Needs neuroradiology + neurosurgery collaboration

Quick-Selection Guide

Full Research Protocol

CT-Based Opportunistic Bone Mineral Density Screening for Undiagnosed Osteoporosis in Patients Undergoing Routine Contrast-Enhanced Abdominal CT

1. TITLE PAGE

2. BACKGROUND AND RATIONALE

• A meta-analysis (Zhu et al., Clinical Imaging, 2021, PMID 34530357) of 10 studies reported pooled sensitivity 0.83 and specificity 0.74 for CT-based detection of osteoporosis using HU thresholds, with AUC 0.84.
• Pickhardt et al. (Radiology, 2013) established normative L1 trabecular HU values across >20,000 adults, with thresholds of 110 HU (>90% specific) and 135 HU (optimal ROC) for osteoporosis.
• A recent Thai study (Chaisen et al., Diagnostics, 2024, PMID 39767207) found HU ≤142 at L1 had 91.9% sensitivity and 48.4% specificity vs DXA in a predominantly female population (mean age 66.3 years).
• Sobecki et al. (Osteoporosis International, 2025, PMID 40493237) confirmed that a 200 mm² circular ROI at the L1 vertebral body centroid on axial imaging is the optimal, reproducible measurement technique, with excellent intra- and inter-observer reliability (ICC >0.95).

3. RESEARCH GAP

4. OBJECTIVES

4.1 Primary Objective

4.2 Secondary Objectives

1. To compare mean L1 HU values across age groups and sex.
2. To identify clinical and demographic predictors of low L1 HU on CT.
3. To determine the proportion of patients with low HU who had no prior DXA or osteoporosis diagnosis documented in medical records.
4. To assess whether HU values differ between non-contrast and contrast-enhanced CT phases (in patients where both are available).
5. To identify the prevalence of incidental vertebral compression fractures in the same cohort.

5. RESEARCH QUESTION (PICO Format)

6. STUDY DESIGN

• Type: Retrospective cross-sectional study
• Setting: Radiology department, [Hospital Name], [Institution]
• Duration: CT scans performed over a 2-year period (e.g., January 2022 - December 2023)
• Ethics: Retrospective use of anonymized imaging and medical record data; minimal risk. Waiver of informed consent to be sought from institutional ethics committee (IEC/IRB).

7. STUDY POPULATION

7.1 Inclusion Criteria

• Adults aged ≥50 years
• Underwent contrast-enhanced abdominal CT (portal venous phase) for any clinical indication other than known osteoporosis or metabolic bone disease
• L1 vertebra adequately visualized on axial images without significant artifact
• Medical records accessible for clinical data extraction

7.2 Exclusion Criteria

• Known pre-existing diagnosis of osteoporosis or metabolic bone disease documented before the CT date
• Presence of metal implants, vertebral augmentation (kyphoplasty/vertebroplasty), or extensive degenerative changes at L1 affecting HU measurement
• Active osseous metastatic disease or primary bone tumor involving the L1 vertebra
• Vertebral fracture at L1 (as this alters trabecular HU)
• Scans with significant motion or streak artifact at L1 level
• Patients who underwent the same CT scan >once (only most recent scan included)
• Severe scoliosis precluding reliable L1 measurement

7.3 Sub-Group of Interest (for secondary analysis)

• Post-menopausal women (aged ≥45 years, documented amenorrhea ≥12 months)
• Patients on chronic corticosteroids (≥5 mg prednisolone equivalent for ≥3 months)
• Oncology patients on aromatase inhibitors or androgen deprivation therapy
• Patients with CKD (eGFR <30 mL/min)

8. SAMPLE SIZE CALCULATION

n = Z² × p(1-p) / d²

• Z = 1.96 (95% confidence level)
• p = expected prevalence of osteoporosis in target population ≈ 0.25 (25%, based on published estimates in adults >50 years; conservative estimate)
• d = acceptable margin of error = 0.05 (±5%)

9. METHODOLOGY

9.1 CT Data Retrieval

• Identify all patients ≥50 years who underwent contrast-enhanced abdominal/abdominopelvic CT during the study period from the PACS/RIS database.
• Anonymize all retrieved CT datasets before analysis.

9.2 L1 HU Measurement - Standardized Protocol

1. Open the CT scan in standard DICOM viewer (e.g., OsiriX, RadiAnt, or PACS workstation).
2. Identify the L1 vertebra as the first non-rib-bearing lumbar vertebra on sagittal scout or reformatted images.
3. Navigate to the axial (transverse) plane at the mid-body level of L1.
4. Place a circular ROI of 200 mm² at the vertebral body centroid, in the anterior trabecular region, avoiding cortical bone, basivertebral vein, osteophytes, and any sclerotic/lytic areas.
5. Record the mean HU value within the ROI.
6. If the axial plane is unsatisfactory, a sagittal anterior ROI placement is acceptable as a secondary option.
7. All measurements performed on soft tissue window settings (W:400, L:40 or W:350, L:35).

9.3 Data Collection Form (per patient)

• Age, sex, BMI, menopausal status (women)
• Ethnicity

• CT indication
• Presence of known diabetes, CKD, liver disease, malignancy
• Current medications: corticosteroids, bisphosphonates, aromatase inhibitors, androgen deprivation therapy, PPIs, anticonvulsants
• Prior DXA scan documented in records (yes/no; result if available)
• Prior osteoporosis or fracture diagnosis (yes/no)
• Referral to endocrinology/rheumatology for bone health (yes/no)

• CT scanner make and model, kVp, mAs
• Contrast phase (non-contrast, arterial, portal venous, delayed)
• L1 HU value
• Presence of vertebral compression fracture at any level (yes/no; level if yes)
• Quality of measurement (adequate/suboptimal/excluded)

9.4 Inter-Observer Reliability

• L1 HU measurements performed independently by two observers (one radiologist with ≥3 years experience; one radiology resident).
• A randomly selected subset of 50 scans (10-15%) measured by both observers.
• Inter-observer agreement assessed using Intraclass Correlation Coefficient (ICC).
• Discordant readings (>15 HU difference) adjudicated by a third senior radiologist.

10. STATISTICAL ANALYSIS PLAN

10.1 Descriptive Statistics

• Continuous variables: mean ± SD or median (IQR) based on normality (Shapiro-Wilk test)
• Categorical variables: frequencies and percentages
• L1 HU distribution plotted as histogram

10.2 Primary Outcome

• Prevalence of low bone mass (HU ≤135) and osteoporosis (HU ≤110) reported as percentage with 95% confidence intervals (Wilson score method)

10.3 Secondary Analyses

• Independent t-test or Mann-Whitney U for HU comparison between sexes
• One-way ANOVA or Kruskal-Wallis for HU comparison across age groups (50-59, 60-69, 70-79, ≥80 years)
• Pearson or Spearman correlation between L1 HU and age, BMI
• Logistic regression (univariate then multivariate) to identify independent predictors of HU ≤110: age, sex, BMI, steroid use, CKD, malignancy, menopausal status
• Sensitivity analysis comparing HU on contrast vs. non-contrast phases (paired t-test in subgroup with both phases available)
• Chi-square test to compare prevalence across subgroups (post-menopausal women vs. men, steroid users vs. non-users)
• ICC with 95% CI for inter-observer agreement (≥0.90 = excellent)

10.4 Reporting Rate Analysis

• What proportion of patients with HU ≤110 had:
  • A prior DXA documented
  • A prior osteoporosis diagnosis
  • An existing referral for bone health assessment

10.5 Software

• SPSS v25 or higher / R version ≥4.2 / MedCalc
• p <0.05 considered statistically significant
• No correction for multiple comparisons in secondary exploratory analyses (reported as hypothesis-generating)

11. ETHICAL CONSIDERATIONS

12. EXPECTED OUTCOMES AND SIGNIFICANCE

• Prevalence of osteoporosis (HU ≤110) expected in ~20-30% of patients aged >50 years based on published data, with higher rates in post-menopausal women and steroid users.
• A substantial proportion (estimated 50-70%) will have had no prior DXA or osteoporosis diagnosis - the "treatment gap."

• Provides local/regional prevalence data for opportunistic screening
• Demonstrates the scale of underdiagnosis using an already-performed, no-additional-cost investigation
• Supports the case for implementing a systematic reporting protocol for low L1 HU in radiology departments
• Directly actionable: could prompt referral for DXA, initiation of bisphosphonate therapy, and fracture prevention

13. LIMITATIONS

1. Contrast enhancement effect: Portal venous phase contrast may elevate HU by ~10-30 HU, potentially underestimating the true prevalence. Addressed via sensitivity analysis and conservative threshold adjustments.
2. Single-center retrospective design: Limits generalizability; selection bias possible.
3. No DXA correlation in all patients: Most patients will not have a DXA for direct correlation, so DXA-validated thresholds are used as proxies.
4. Scanner variability: HU values can vary between CT scanners; scanner model and parameters will be recorded as covariates.
5. L1-only measurement: Lumbar spine DXA typically averages L1-L4; single vertebra measurement may not reflect overall spinal BMD equally well in all patients.
6. Medical record completeness: Underreporting of prior DXA or medications possible in retrospective records.

14. TIMELINE

15. PUBLICATION TARGETS

16. KEY REFERENCES

1. Pickhardt PJ, et al. CT-based screening for osteoporosis at routine abdominal and thoracic CT. Radiology. 2013;268(1):120-127.
2. Zhu Y, Triphuridet N, Yip R, et al. Opportunistic CT screening of osteoporosis on thoracic and lumbar spine: a meta-analysis. Clin Imaging. 2021;80:111-122. [PMID: 34530357]
3. Chaisen M, Sritara C, Chitrapazt N, et al. Opportunistic Screening for Osteoporosis by CT as Compared with DXA. Diagnostics (Basel). 2024;14(24):2846. [PMID: 39767207]
4. Sobecki JN, Krueger D, Pickhardt PJ, et al. Optimizing region of interest size and placement for clinical opportunistic CT trabecular bone Hounsfield unit measurements. Osteoporos Int. 2025. [PMID: 40493237]
5. Guenoun D, Champsaur P. Opportunistic Computed Tomography Screening for Osteoporosis and Fracture. Semin Musculoskelet Radiol. 2023;27(4). [PMID: 37748468]
6. Deshpande N, Hadi MS, Lillard JC, et al. Alternatives to DEXA for the assessment of bone density: a systematic review. J Neurosurg Spine. 2023;38(4). [PMID: 36609369]
7. Reschke P, Koch V, Sommer CM, et al. The Association between opportunistic DECT-derived bone values and vertebral fracture status. Eur J Radiol. 2025. [PMID: 40373671]

17. TEAM ROLES (Suggested)