2 | Scientific Reports

2

From: Development of an AI-based magnetic resonance imaging reading support program (AMP) for deep endometriosis diagnosis

2

Applying the nnU-Net-based model to evaluate plaque location and thickness. a The ground truth and b model prediction for the segmentation masks of the plaque and four organs which include the uterus (blue), bladder (light blue), rectum (green)*, ovary (yellow) and plaque (red). c The magnified view of b focusing on the segmentation of the plaque (red). The plaque depth measured from the posterior surface of the uterus and its perpendicular width are shown in white and blue, respectively. d The relationship between the automatically measured plaque depth based on the model’s segmentation mask and the plaque depth directly measured from the MR image by an expert radiologist. * In this study, the sigmoid colon was included in the “rectum” as a target area in some cases.

Back to article page