Clinical Scorecard: Ultrasound-Specific Augmentation for Enhanced Segmentation of Ovarian Tumors in B-Mode and Contrast-Enhanced Ultrasound Images

At a Glance

Key Highlights

• Combined augmentation improved segmentation metrics for contrast-enhanced ultrasound.
• Dice coefficient for CEUS improved from 0.468 to 0.476.
• 62.9% of cases showed improvement in segmentation performance.
• The proportion of CEUS cases with Dice below 0.5 decreased from 55.9% to 51.8%.
• Training-only augmentation strategy supports modality-shift robustness.

Guideline-Based Recommendations

Diagnosis

• Utilize standardized terminology and structured ultrasound features for adnexal masses.

Management

• Incorporate segmentation models to support size measurement and shape analysis.

Monitoring & Follow-up

• Evaluate segmentation performance across different ultrasound modes.

Risks

• Segmentation models may learn appearance patterns specific to acquisition settings.

Patient & Prescribing Data

Patients undergoing ultrasound evaluation for ovarian tumors.

Segmentation aids in risk assessment and treatment planning.

Clinical Best Practices

• Implement ultrasound-specific augmentation to enhance segmentation robustness.
• Ensure consistent visual interpretation and reproducible feature extraction.

Related Resources & Content

• International Ovarian Tumor Analysis framework
• ADNEX model
• O-RADS ultrasound