Abstract | Medical imaging technology has rapidly advanced in the last few decades, providing detailed images of the human body. The accurate analysis of these images and the segmentation of anatomical structures can produce significant morphological information, provide additional guidance toward subject stratification after diagnosis or before a clinical trial, and help predict a medical condition. Usually, medical scans are manually segmented by expert operators, such as radiologists and radiographers, which is complex, time-consuming and prone to inter-observer variability. A system that generates automatic, accurate quantitative organ segmentation on a large scale could deliver a clinical impact, supporting current investigations in subjects with medical conditions and aiding early diagnosis and treatment planning. This paper proposes a web-based application that automatically segments multiple abdominal organs and muscle, produces respective 3D reconstructions and extracts valuable biomarkers using a deep learning backend engine. Furthermore, it is possible to upload image data and access the medical image segmentation tool without installation using any device connected to the Internet. The final aim is to deliver a web- based image-processing service that clinical experts, researchers and users can seamlessly access through IoT devices without requiring knowledge of the underpinning technology. |
---|