biostudies-literatureZhang Jthe Natural Science Foundation of Chongqingthe open project of Chongqing Research Center for Prevention & Control of Maternal and Child Diseases and Public Healththe open project of Chongqing Research Center for Prevention & Control of Maternal and Child Diseases and Public Health3115https://www.ebi.ac.uk/biostudies/studies/S-EPMC12830614biostudies-literatureUnknown16(1)To overcome inter-observer variability in conventional antral follicle count (AFC) assessment and AMH testing limitations, we developed an AI-powered framework using routine 2D ultrasound to standardize ovarian-reserve evaluation in assisted reproductive technology (ART). This multicenter retrospective study analyzed 395 women with infertility from two affiliated hospitals. The cohort was divided into training (n = 210), internal-test (n = 91), and external-test (n = 94) cohorts. We established three prediction models: radiomics model, 674 IBSI-compliant features; deep-learning model, ResNet50-based feature extraction; fusion model, hybrid approach combining both modalities. Model performance was validated against the manual AFC and serum AMH levels. Sequential classification categorized pScientific reportsUltrasound radiomics and deep learning for predicting antral follicle count and anti-Mullerian hormone.PMC12830614CQFYSJ01002CSTB2023NSCQ-BHX0100Rong YRan SZhang JZhong CNi DLiu SfalseUltrasound radiomics and deep learning for predicting antral follicle count and anti-Mullerian hormone.To overcome inter-observer variability in conventional antral follicle count (AFC) assessment and AMH testing limitations, we developed an AI-powered framework using routine 2D ultrasound to standardize ovarian-reserve evaluation in assisted reproductive technology (ART). This multicenter retrospective study analyzed 395 women with infertility from two affiliated hospitals. The cohort was divided into training (n = 210), internal-test (n = 91), and external-test (n = 94) cohorts. We established three prediction models: radiomics model, 674 IBSI-compliant features; deep-learning model, ResNet50-based feature extraction; fusion model, hybrid approach combining both modalities. Model performance was validated against the manual AFC and serum AMH levels. Sequential classification categorized p2025-01-01T00:00:00Z2025 Dec2026-06-06T20:10:13.173Z2026-06-04T03:14:18.704ZS-EPMC128306144142229710.1038/s41598-025-33010-w