Role of PET-CT in Obstetrics and Gynaecology

Introduction

Principles of FDG-PET/CT

• FDG (a glucose analogue) is taken up preferentially by metabolically active (malignant) cells
• It is phosphorylated to FDG-6-phosphate and trapped inside cells
• Increased uptake is quantified using the Standardised Uptake Value (SUV); SUVmax is the key metric
• Image fusion with CT provides both metabolic and anatomical localisation
• Spatial resolution limitation: false negatives occur with lesions smaller than 0.5 cm
• False positives: inflammatory/infective lesions, post-operative changes, reactive lymphadenopathy, granulomatous disease

Applications in Gynaecological Oncology

1. Cervical Cancer

• FDG-PET/CT is particularly useful in advanced cervical cancer (FIGO IIb-IVb) for detecting lymph node metastases, especially at unexpected sites beyond the pelvis and retroperitoneum (e.g., supraclavicular nodes)
• It is better than conventional cross-sectional imaging for lymph node detection - conventional CT/MRI rely on size criteria and miss microscopic disease; PET detects metabolic activity regardless of node size
• Sensitivity: 75%-100%, Specificity: 87%-100% for lymph node metastases
• Although lymphadenopathy is not formally part of FIGO staging, detection of distant nodal disease (above renal hila, inguinal) classifies as stage IVb and directly alters treatment planning
• Value in early-stage disease (FIGO I-IIa) is limited and not routinely recommended

• SUVmax of primary tumour at diagnosis is a sensitive biomarker of treatment response and prognosis (Kidd et al.)
• SUV of pelvic lymph nodes predicts treatment response, pelvic recurrence risk, and disease-specific survival

• PET/CT is highly sensitive in detecting recurrent cervical cancer, particularly pelvic, para-aortic, and distant nodal recurrence
• Clinically important when CA-125 rises without identifiable lesion on CT/MRI

• Used to define the gross tumour volume (GTV) for conformal radiotherapy and brachytherapy planning
• Guides extended-field radiotherapy when para-aortic nodal involvement is confirmed

2. Endometrial Cancer

• Primary staging is surgical; MRI is the imaging modality of choice for local staging (myometrial invasion, cervical extension)
• FDG-PET/CT plays a role in high-risk or advanced endometrial cancer (type 2: serous papillary, clear-cell carcinomas) for detecting distant metastases and lymph node spread
• Used for detection of recurrence - PET/CT detects recurrent endometrial cancer earlier than CT alone, particularly in lymph nodes and peritoneum
• Recent literature (Causa Andrieu et al., 2025, PMID: 40835435) endorses PET/CT in the management algorithm for endometrial cancer especially in advanced and recurrent settings
• Not routinely indicated for early-stage (FIGO I) low-grade endometrioid carcinoma

3. Ovarian Cancer

• CT remains the primary modality for staging, surgical planning, and follow-up of ovarian cancer (extent of peritoneal disease guides treatment selection)
• FDG-PET/CT roles:
  • Detection of recurrent ovarian cancer - especially useful when CA-125 elevation is not explained by CT findings; PET identifies metabolically active peritoneal implants and nodal disease
  • Distinguishing viable tumour from post-treatment fibrosis after chemotherapy
  • Assessing response to chemotherapy
  • Detecting distant metastases (supradiaphragmatic, mediastinal, distant)
• Limitation: low-grade serous and mucinous tumours show low FDG avidity - false negatives are common

4. Vulvar and Vaginal Cancer

• FDG-PET/CT is used for inguinal and pelvic lymph node staging
• Identifies nodal metastases that guide the decision for sentinel lymph node biopsy vs. inguino-femoral lymphadenectomy
• Useful for detecting recurrent vulvar cancer

5. Gestational Trophoblastic Disease (GTD)

• PET/CT has a limited but emerging role in high-risk gestational trophoblastic neoplasia (GTN) resistant to chemotherapy
• Helps identify viable disease and sites of resistance to guide salvage therapy
• Generally avoided in low-risk GTN due to radiation exposure concerns

6. Radiotherapy Planning

• Across all gynaecological malignancies, PET/CT improves target volume delineation for external beam radiotherapy
• PET-guided brachytherapy helps avoid geographic miss and reduces dose to organs at risk

Role in Obstetrics

• Ionising radiation exposure to the fetus (CT component)
• FDG transfer across the placenta with theoretical fetal risk
• In cases of cancer diagnosed during pregnancy, PET/CT may be considered post-partum or, in exceptional circumstances, in the third trimester when maternal benefit clearly outweighs fetal risk
• Staging of cervical cancer in pregnancy typically defers PET/CT until after delivery

Advantages and Limitations

Summary

1. Staging advanced cervical cancer - especially nodal mapping beyond conventional imaging
2. Detecting recurrence of cervical, ovarian, and endometrial cancers
3. Radiotherapy planning - target volume delineation
4. Prognostication via SUVmax metrics

Role of MRI in Obstetrics and Gynaecology

Introduction

Technical Basis and Sequences Used

Role in Gynaecology

1. Endometrial Cancer

• Determines the depth of myometrial invasion (FIGO Ia vs Ib) - the most critical prognostic factor
  • Stage Ia: tumour signal extends to <50% myometrial depth
  • Stage Ib: tumour signal extends to ≥50% myometrial depth
• Detects cervical stromal invasion (Stage II) - disruption of low-signal cervical stroma by tumour
• Differentiates endometrial vs cervical primary tumour when biopsy is inconclusive
• Identifies adnexal extension (Stage IIIa) and vaginal/parametrial involvement (Stage IIIb)
• Detects lymph node metastases - nodes >1 cm in short axis, with abnormal signal, necrosis, or irregular contour (Stage IIIc)
• Detects bladder or bowel mucosal invasion (Stage IVa)
• DWI is particularly useful when the tumour is iso- or hyperintense to myometrium on T2, improving myometrial invasion accuracy and replacing gadolinium when contrast is contraindicated
• Enables risk stratification - guides selection of patients for lymph node dissection, thereby avoiding unnecessary extended surgery in low-risk disease

2. Cervical Cancer

• Tumour size and location (exophytic vs endocervical) - measured on T2-weighted images
• Stromal invasion depth - low-signal stromal ring disruption indicates Stage IB or above
• Parametrial invasion (Stage IIb) - spiculated tumour-parametrium interface on axial oblique T2; high accuracy avoids unnecessary surgery
• Vaginal extension - disruption of low-signal vaginal wall (upper two-thirds = Stage IIa; lower third = Stage IIIa)
• Pelvic side-wall invasion (Stage IIIb) - tumour extending to iliac vessels or pelvic musculature
• Bladder/rectal invasion (Stage IVa) - disruption of normal mucosal low-signal intensity
• In fertility-sparing candidates (trachelectomy), MRI confirms: tumour <2 cm, distance from internal os ≥1 cm, no parametrial invasion
• DCE-MRI differentiates tumour recurrence from post-radiation fibrosis
• Radiotherapy planning - target volume delineation for brachytherapy and EBRT

3. Ovarian Cancer

• MRI is a problem-solving tool for indeterminate adnexal masses on ultrasound - tissue-specific signal characteristics on T1/T2 sequences characterise:
  • Endometrioma: high signal on T1, "shading" on T2
  • Mature teratoma (dermoid): fat signal on T1, suppressed on fat-sat
  • Mucinous vs serous tumours
  • Clear-cell carcinoma: restricted diffusion on DWI confirms malignant mural nodules
• Best technique for pelvic side-wall invasion assessment in ovarian cancer
• Assesses resectability of solitary pelvic recurrences - determines whether secondary cytoreduction is feasible
• Identifies serosal/peritoneal deposits that are equivocal on CT

4. Uterine Fibroids (Leiomyomata) and Adenomyosis

• MRI is the most accurate technique for mapping fibroid size, number, and location (submucosal, intramural, subserosal, pedunculated) - essential for surgical planning (myomectomy vs hysterectomy) and uterine artery embolisation (UAE)
• Pre-UAE MRI confirms fibroid vascularity and excludes adenomyosis (which responds poorly to UAE)
• Adenomyosis: thickened junctional zone >12 mm on T2-weighted imaging; high T1 foci within myometrium

5. Adnexal Masses - Characterisation

• MRI achieves high specificity (>90%) for characterisation of adnexal masses that are indeterminate on ultrasound
• The O-RADS MRI reporting system (analogous to IOTA/PI-RADS) standardises reporting

6. Endometriosis and Deep Infiltrating Endometriosis (DIE)

• MRI detects deep infiltrating endometriosis involving bowel (rectosigmoid), bladder, ureters, uterosacral ligaments, and ovaries
• Essential for surgical planning before laparoscopic DIE excision
• T1 fat-suppressed sequences: haemorrhagic endometriotic deposits appear as high-signal foci

7. Vulvar and Vaginal Cancer

• MRI defines local tumour extent, vaginal wall and urethral involvement
• Identifies inguinal lymph node enlargement
• Critical for radiotherapy planning

Role in Obstetrics

1. Adjunct to Ultrasound for Fetal Anomalies

2. Placental Disorders

• Placenta accreta spectrum (PAS) - MRI is the best investigation after USS for confirming and grading accreta/increta/percreta:
  • Dark intraplacental bands on T2 ("placental lakes")
  • Uterine bulge, bladder wall invasion in percreta
  • Critical for surgical planning (MDT approach, planned caesarean hysterectomy)
• Placenta praevia - MRI defines relationship to internal os and cervix when USS is inconclusive
• Vasa praevia - MRI with HASTE sequences can confirm fetal vessel position

3. Suspected Appendicitis or Other Acute Abdominal Pathology in Pregnancy

• MRI without gadolinium is the preferred cross-sectional modality for pregnant patients with suspected appendicitis when USS is equivocal - avoids radiation of CT

4. Maternal Pelvic Pathology in Pregnancy

• Assessment of ovarian masses complicating pregnancy (persistence beyond 1st trimester)
• Assessment of uterine fibroids causing complications
• Evaluation of Mullerian anomalies - bicornuate, septate, or unicornuate uterus affecting obstetric outcome

5. Fetal Growth Restriction / Brain Maturation

• MRI can assess fetal brain maturation and white matter injury in suspected intrauterine growth restriction when neonatal outcome counselling is required

Advantages and Limitations

Summary

1. Local staging of endometrial cancer - myometrial invasion depth and risk stratification
2. Local staging of cervical cancer - parametrial invasion, determining surgery vs chemoradiotherapy
3. Problem-solving for adnexal masses - characterising indeterminate lesions after USS
4. Deep infiltrating endometriosis - surgical planning
5. Fibroid mapping and UAE planning

• Fetal brain and CNS anomaly characterisation
• Placenta accreta spectrum assessment
• Acute abdominal pathology in pregnancy (appendicitis, ovarian pathology)

Role of Artificial Intelligence in Obstetrics and Gynaecology

Introduction

Key AI Technologies Relevant to O&G

Applications in Obstetrics

1. Prenatal Imaging and Fetal Anomaly Detection

• AI-powered ultrasound systems use deep learning (CNN) to:
  • Automatically measure fetal biometry (BPD, HC, AC, FL) with expert-level accuracy
  • Detect congenital anomalies (cardiac defects, neural tube defects, cleft lip/palate) in real-time
  • Standardise image acquisition - AI guides the sonographer to correct planes, reducing operator dependence
  • Perform fetal organ segmentation and automated growth chart plotting
• These tools are especially valuable in low-resource settings where trained sonographers are scarce
• Example: AI systems demonstrated sensitivity >85% for detecting major structural anomalies, comparable to experienced perinatologists

2. Fetal Monitoring - Cardiotocography (CTG) Interpretation

• CTG interpretation is notoriously subjective with high inter-observer variability
• AI algorithms (ML-based) analyse CTG traces to:
  • Classify fetal heart rate patterns (reassuring, suspicious, pathological) more objectively
  • Predict fetal distress and hypoxia earlier than manual interpretation
  • Reduce unnecessary interventions while avoiding missed pathology
• Systems like Dawes-Redman criteria embedded in computerised CTG (cCTG) analysis represent early AI integration
• Deep learning models trained on thousands of labelled CTG traces now outperform junior clinicians in pattern recognition

3. Preterm Birth Prediction

• Preterm birth (<37 weeks) is a leading cause of neonatal morbidity
• ML models integrating cervical length, clinical history, biomarkers (fFN, PAMG-1), demographic data, and EHR variables outperform traditional single-parameter screening
• AI-based prediction models achieve AUC >0.85 for spontaneous preterm birth risk
• Enables targeted prophylactic interventions (progesterone, cervical cerclage) in high-risk women

4. Pre-eclampsia and Hypertensive Disorders Prediction

• AI models integrating uterine artery Doppler indices, MAP, PAPP-A, PlGF, clinical risk factors (FMF algorithm) in the first trimester predict pre-eclampsia with high sensitivity
• Deep learning on EHR data can predict late-onset pre-eclampsia (after 34 weeks) which is harder to screen conventionally
• AI surveillance tools alert clinicians to rising blood pressure trends before clinical diagnosis (Zapata et al., 2025, [PMID: 40202855])

5. Gestational Diabetes Mellitus (GDM) Prediction

• ML algorithms trained on booking blood glucose, BMI, ethnicity, family history, and previous obstetric history predict GDM risk before the diagnostic OGTT
• Allows early lifestyle intervention and monitoring in high-risk groups

6. Postpartum Haemorrhage (PPH) Risk Prediction

• AI tools analyse intrapartum factors (labour duration, oxytocin use, uterine atony predictors) to stratify PPH risk prospectively
• Enables preoperative planning (cell salvage, blood availability, uterotonic protocols) in real-time

7. Placenta Accreta Spectrum (PAS) Diagnosis

• CNN-based models applied to ultrasound and MRI images identify placental lacunae, loss of retroplacental clear space, and bladder wall invasion with high accuracy
• Supports earlier diagnosis and MDT surgical planning

Applications in Gynaecology

1. Cervical Cancer Screening and Colposcopy

• AI-assisted colposcopy (deep learning on colposcopic images) classifies cervical lesions as normal, low-grade, or high-grade with accuracy comparable to expert colposcopists
• AI-based cervical cytology (Pap smear): CNN algorithms screen liquid-based cytology slides, flagging abnormal cells and reducing false-negative rates - near-human accuracy in cervical cancer detection
• HPV subtyping combined with AI risk stratification guides triage (surveillance vs immediate colposcopy vs treatment)
• Especially impactful in countries lacking cytopathology workforce

2. Gynaecological Oncology

• ML models integrating CA-125 trends, clinical data, and ultrasound features (IOTA rules, O-RADS) improve risk of malignancy classification beyond RMI
• AI analysis of CT/MRI images predicts surgical resectability and response to chemotherapy

• AI applied to histopathology slides (digital pathology) classifies tumour grade, FIGO stage, and molecular subtypes (POLE mutant, MSI, CNH, NSMP) from H&E slides alone - potential to replace expensive molecular testing
• Mondal et al. (2026, [PMID: 42366266]) reviewed AI integration with multi-omics data for precision oncology in endometrial cancer

• AI-assisted PET/CT image analysis improves lymph node staging accuracy and target volume delineation for radiotherapy

3. Endometriosis

• Endometriosis is notoriously difficult to diagnose (average delay ~7-10 years)
• ML models combining symptom profiles, menstrual history, biomarkers, and imaging features improve pre-surgical diagnosis
• AI analysis of laparoscopic video identifies endometriotic lesions in real-time during surgery
• AI tools in emergency settings demonstrate superior performance vs clinician estimation in endometriosis and acute pelvic pain (Elbiss & Abu-Zidan, 2025, [PMID: 39924503])

4. Assisted Reproductive Technology (ART) / IVF

• AI-based embryo selection is one of the most validated AI applications in O&G
  • Time-lapse imaging + CNN analysis evaluates embryo morphokinetics objectively
  • AI selects the embryo with highest implantation potential, outperforming conventional morphological grading by embryologists
  • Reduces multiple embryo transfer and improves single-embryo transfer success rates
• Ovarian stimulation protocol optimisation - ML predicts optimal gonadotrophin dose based on patient's ovarian reserve markers (AMH, AFC, FSH)
• Sperm morphology analysis - AI automated assessment replaces subjective manual WHO criteria

5. Robotic Surgery Enhancement

• AI-integrated robotic platforms (e.g., da Vinci Xi + AI modules) assist in:
  • Tissue recognition and haptic feedback simulation
  • Identifying critical anatomical structures (ureters, vessels) to prevent iatrogenic injury
  • Tracking surgical gestures and providing real-time performance feedback
  • Autonomous suturing algorithms in development
• AI analyses surgical video to assess technical skill and provide training feedback (surgical education)

6. Ovarian Reserve Assessment and PCOS Diagnosis

• AI algorithms analysing antral follicle count from 3D ultrasound provide more accurate ovarian reserve estimation than manual 2D counts
• ML models combining clinical, biochemical (LH:FSH ratio, AMH, testosterone), and ultrasound data improve PCOS diagnosis accuracy beyond Rotterdam criteria alone

AI in Clinical Decision Support

• Ingest EHR data, diagnostics, and clinical guidelines
• Provide real-time risk alerts (e.g., flagging women at high risk for GDM, preterm birth, VTE)
• Guide triage decisions in emergency O&G (acute abdomen, ectopic pregnancy risk scores)
• Support treatment selection (chemo protocol, surgical approach) based on patient-specific data
• A systematic review identified 30 studies of AI-augmented CDSS in pregnancy, covering prenatal care through postpartum

Advantages

1. Consistency - no fatigue, bias, or inter-observer variability
2. Speed - instant analysis of complex imaging or data
3. Pattern recognition - detects subtle patterns invisible to human observers
4. Integration - synthesises multimodal data (imaging + labs + clinical history)
5. Equity - extends specialist-level expertise to low-resource settings
6. Predictive power - identifies risk before clinical symptoms manifest

Challenges and Limitations

Ethical Considerations

• AI must be developed with diverse, representative training datasets to avoid amplifying health disparities
• Explainability (XAI) frameworks should be incorporated so clinicians understand AI reasoning
• AI functions as a decision-support tool, not a replacement for clinical judgement
• Human supervision remains mandatory - AI output must be verified against clinical context
• Patient informed consent for AI-assisted care is an emerging requirement

Summary

Fetal MRI

Introduction

Technical Basis

Advantages Over Ultrasound

Indications

1. Central Nervous System (CNS) Anomalies

• Defined as lateral ventricle atrial diameter ≥10 mm
  • Mild: 10-12 mm; Moderate: 12-15 mm; Severe: >15 mm
• MRI is recommended by the Society for Maternal-Fetal Medicine (SMFM) when dedicated neurosonography is unavailable or equivocal
• MRI readily differentiates hydranencephaly (absent supratentorial tissue - lethal) from aqueductal stenosis with thinned but present cerebrum (amenable to postnatal shunting) - a distinction that directly alters prognosis and management
• A 2020 international multicentre study supported MRI assessment in every fetus with ventriculomegaly; if isolated on neurosonography, MRI reassures parents of low associated anomaly risk
• Idiopathic mild ventriculomegaly (10-12 mm, isolated, male sex, >20 weeks) has a good outcome - but MRI is required to exclude subtle co-existing abnormalities

• Colpocephaly (teardrop-shaped dilated atria), absent cavum septum pellucidum, elevated third ventricle are USS clues
• MRI confirms complete vs partial agenesis and identifies cortical and posterior fossa anomalies which are the most commonly missed findings at USS alone (meta-analysis 2021)
• Schizencephaly - MRI shows low-signal grey matter lining the wedge-shaped cleft (porencephaly has no grey matter lining)

• Lissencephaly (smooth, agyric brain), Pachygyria (broad flat gyri), Polymicrogyria, Grey matter heterotopia (arrested neuronal migration)
• These cannot be detected on USS but are clearly shown on MRI by their signal intensity characteristics
• MRI differentiates subependymal heterotopic nodules (follow grey matter signal) from tuberous sclerosis tubers (do not follow grey matter signal)

• Dandy-Walker malformation, Blake's pouch cyst, mega cisterna magna, cerebellar hypoplasia - MRI defines the vermis and cerebellar hemispheres far better than USS, directly impacting prognosis counselling

• Lobar HPE in particular benefits from fetal MRI to define residual grey matter and associated cortical anomalies

• T1WI detects haemorrhage (high signal) with corresponding T2 low signal
• DWI demonstrates diffuse ischaemic injury earlier than USS - critical for delivery planning and medico-legal documentation that injury did not occur intrapartum
• Progressive ventriculomegaly following hypoxic-ischaemic injury (periventricular leukomalacia) is quantifiable on MRI

2. Congenital Diaphragmatic Hernia (CDH)

• MRI is superior to USS for:
  • Fetal lung volume (FLV) measurement - total and observed/expected (O/E) lung-to-head ratio using volumetric sequences
  • Determining liver position (liver-up vs liver-down) - a key prognostic marker for survival
  • O/E total FLV <25% with liver herniation = severe CDH; associated with high mortality and pulmonary hypertension risk
• FLV on MRI correlates with survival, postnatal course severity, and long-term morbidity (Amodeo et al., 2022, [PMID: 35794403])
• Guides decision for fetal endoscopic tracheal occlusion (FETO) therapy in severe left-sided CDH
• MRI identifies herniated organs (stomach, bowel, spleen, liver) with anatomical precision

3. Fetal Abdominal and Chest Pathology

• MRI measures lesion volume and calculates CVR (CPAM volume ratio) to predict hydrops risk
• Differentiates CPAM from sequestration by identifying feeding vessel on MRI

• Differentiates gastroschisis from omphalocele with greater certainty
• Assesses bowel quality (echogenicity, dilatation) in gastroschisis - MRI shows bowel wall thickness and content

• MRI delineates intrapelvic/presacral extension (Altman classification) which is critical for planning surgical approach
• Assesses vascularity and solid vs cystic components

• Identifies bowel signal characteristics: meconium (high T1 signal in normal terminal ileum vs absent signal in jejunal/ileal atresia)
• Detects Hirschsprung disease - absence of normal meconium signal in rectum

4. Genitourinary Anomalies

• Differentiates cloacal anomaly from other urogenital malformations
• Defines anatomy of complex urogenital sinus abnormalities for surgical planning
• Quantifies renal volume in obstructive uropathy and bilateral renal agenesis (lethal vs non-lethal scenarios)

5. Fetal Neck and Airway Masses

• Cervical teratoma / lymphatic malformation (cystic hygroma)
• MRI defines airway involvement, tongue/floor of mouth extension, and mediastinal extension
• Critical for planning EXIT (Ex-Utero Intrapartum Treatment) procedure - airway secured while fetal circulation is maintained
• MRI determines feasibility and timing of EXIT procedure to prevent neonatal asphyxia

6. Vascular and Placental Indications

• Placenta Accreta Spectrum (PAS): MRI is performed after inconclusive USS to confirm and grade PAS
  • Dark T2 intraplacental bands, uterine bulge, bladder wall invasion (percreta)
  • Guides planning of MDT caesarean hysterectomy
• Vasa praevia - HASTE sequences confirm vessel position in relation to cervical os
• Twin-twin transfusion syndrome (TTTS) - MRI assesses brain injury in both twins prior to laser ablation

7. Fetal Brain Maturation Assessment

• Sulcation and gyration progress in a predictable, gestational-age-dependent sequence
• MRI (ideally 28-34 weeks) detects delayed cortical maturation - a prognostic marker in preterm infants and IUGR
• Reference normative atlases (Boston fetal brain atlas) allow systematic comparison of sulcal development
• Myelination (T1 bright, T2 dark) can be tracked from 28 weeks in internal capsule and brainstem

8. Fetal Tumours

• Brain tumours (e.g., giant-cell astrocytoma in tuberous sclerosis, congenital GBM) - MRI characterises mass, oedema, haemorrhage, and ventricular compression
• Fetal hepatic tumours (haemangioendothelioma, hepatoblastoma) - MRI differentiates lesion type and guides antenatal counselling

Limitations

Impact on Clinical Management

1. Confirms or refutes USS diagnosis - reduces parental anxiety when MRI is normal
2. Identifies additional anomalies not seen on USS (especially CNS) - may upgrade prognosis
3. Guides decision for termination of pregnancy where legal
4. Identifies candidates for in-utero intervention (FETO, EXIT, TTTS laser)
5. Facilitates delivery planning at appropriate centre (NICU, neurosurgery, cardiac surgery on standby)
6. Provides baseline for postnatal management planning (shunting, surgical repair)

Summary

1. CNS anomalies - ventriculomegaly, ACC, cortical malformations, posterior fossa lesions
2. CDH - lung volume measurement and prognosis
3. Airway masses - EXIT procedure planning
4. Placenta accreta spectrum - surgical planning