FADA: Accessible fetal ultrasound interpretation and annotation with a selectively distilled unified vision-language model

2026-06-09

AI Summary

This study addresses the critical challenge of limited access to prenatal ultrasound screening, particularly in low- and middle-income countries, where existing deep learning approaches often rely on multiple models and expert annotations. The authors propose FADA, a unified vision–language model based on Qwen-VL, which integrates clinical interpretation, classification, detection, and segmentation of fetal ultrasound images within a single “interpret-then-annotate” pipeline—eliminating the need for external labels. A novel selective knowledge distillation strategy aligns features only for annotation tasks while applying standard fine-tuning for interpretation, enabling training on a single consumer-grade GPU and offline deployment on edge devices. Evaluated on 237 expert-validated cases, FADA-SKD achieves strong performance in segmentation (Dice = 0.8820), detection (mAP@0.50 = 0.7671), and structured interpretation (100% compliance), with 73.5% of interpretations scoring full marks and end-to-end inference running offline on a Snapdragon 7 Gen 1 mobile device in approximately 60 seconds.

Abstract

A global shortage of trained sonographers limits prenatal ultrasound screening in low- and middle-income countries, where over half of pregnant women receive no skilled sonography. Current deep learning approaches address detection, segmentation, or classification in isolation, each demanding a separate model and expert-specified labels at inference. We present FADA, a unified vision-language model built on Qwen3.5-VL that performs clinical interpretation, classification, detection, and segmentation through a single interpretation-first pipeline without external labels. FADA distills knowledge from four domain-specific foundation models (FetalCLIP, UltraSAM, USF-MAE, UltraFedFM) via offline pre-computed feature caching. Selective distillation, which applies feature alignment only to annotation tasks while interpretation relies on standard fine-tuning, consistently outperforms full distillation across most evaluation axes. The recommended variant, FADA-SKD, achieves 0.8820 mean Dice for segmentation, 0.7671 mAP@0.50 for detection, and 100% structured interpretation compliance. Expert sonographer validation across 237 images confirms clinically acceptable outputs in both autonomous and human-in-the-loop modes, with 73.5% of interpretations scoring perfectly under clinician guidance. The system is trainable on a single consumer GPU and deployable without cloud connectivity. We validate edge deployment by running the compressed 0.8B model on a commodity smartphone (Qualcomm Snapdragon 7 Gen 1, 12 GB RAM) using llama.cpp with GGUF quantization, completing the full 5-phase pipeline in approximately 60 seconds entirely offline. This establishes a practical pathway for integrating AI-assisted fetal assessment with portable ultrasound devices, directly addressing diagnostic access gaps in resource-constrained settings. Code, models, and data are available at https://github.com/mahmoodphd/FADA.

Problem

• fetal ultrasound
• diagnostic access gap
• sonographer shortage
• low-resource settings
• prenatal screening

Innovation

• unified vision-language model
• selective knowledge distillation
• offline edge deployment
• fetal ultrasound interpretation
• annotation-free inference