Researchers have developed AnatoProto, a novel framework designed to improve the detection of standard planes in fetal ultrasound images, particularly in challenging low-cost blind sweeps. This method adapts a frozen BiomedCLIP encoder by incorporating anatomy-weighted spatial pooling and a within-case prototype loss, which leverages case-level structure to enhance frame embeddings. The framework also includes a three-stage cascade refinement and a hybrid prediction head to reduce false positives, achieving a test F1 score of 67.72 on the ACOUSLIC-AI benchmark, significantly outperforming existing baselines. AI
IMPACT Enhances diagnostic accuracy in fetal ultrasound imaging, potentially improving prenatal care.
RANK_REASON Academic paper detailing a novel framework and its performance on a benchmark. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
