OP15.04: Novel AI‐guided ultrasound screening system for fetal heart can demonstrate findings in timeline diagram

• DOI: 10.1002/uog.20796
• Publication Date: 2019-10-01T05:47:16Z
• Authors: R. Komatsu; R. Matsuoka; T. Arakaki; M. Tokunaka; M. Komatsu; A. Sakai; S. Yasutomi; R. Hamamoto; A. Sekizawa
• Keywords: 03 medical and health sciences; 0302 clinical medicine

We have developed a novel fetal cardiac screening system that is used deep leaning technology. This can real-timely and automatically detect each part in echocardiographic movie was acquired 2nd trimester examination. Artificial intelligence (AI) based on convolutional neural networks. Whereas, the decision process of AI, especially learning, has been unexplainable still black box. Therefore, we tried to visualise explain its toward practical utilisation our clinic. 42 normal heart movies were as training dataset, experts had annotated correct positions 18 different parts peripheral organs such atrium, ventricle, blood vessels stomach. The tetralogy Fallot (TOF) transposition great arteries (TGA) test dataset. Each includes sequential axial views from level stomach base heart. AI described barcode-used diagram along with sweep scanning timeline. demonstrated detection probability moment. clearly showed abdomen part, structural out flow tract/blood vessel like observed movie. In case TOF, pulmonary artery not TGA, pattern ones. Supporting information be found online version this abstract Please note: publisher responsible for content or functionality any supporting supplied by authors. Any queries (other than missing content) should directed corresponding author article.