Towards Reliable and Reproducible Fetal Brain Biometry: A Deep Learning Approach Using MRI

作者：Francesca Maccarone, Marina Di Stefano, Giorgio Longari, Giulia Frigerio, Gloria Rizzato, Rocco Prudentino, Nivedita Agarwal, Tommaso Ciceri, Denis Peruzzo, Simone Melzi · 单位：a NeuroImaging Laboratory, Scientific Institute IRCCS Eugenio Medea, Bosisio Parini (LC), Italy, b University of Milano-, Bicocca, Milan, Italy, c Image Sciences Institute, Division Imaging and Oncology, University Medical Center Utrecht, convolutional neural network is trained to regress landmark coordinates from brain segmentation label maps, followed by, evaluated on two publicly available fetal MRI datasets comprising 150 volumes (gestational age range: 20–37 weeks), available automated pipeline, the proposed method achieves comparable or improved accuracy for most measurements., scalability that support integration into clinical workflows., process is time-consuming, labour-intensive, and prone to inter- and intra-observer variability [2,11,12]. Image quality is, including reliance on 2D slice-based analysis, lack of standardized pipelines, limited availability of publicly accessible · 会议/期刊：arXiv preprint · 方向：cs.CV · 发布日期：2026-08-05