Medical Image Segmentation via Unsupervised Convolutional Neural Network

• DOI: 10.48550/arxiv.2001.10155
• Publication Date: 2020-01-01
• Authors: Junyu Chen; Eric C. Frey
• Keywords: FOS: Computer and information sciences; Computer Science - Machine Learning; 03 medical and health sciences; 0302 clinical medicine; Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV); Computer Science - Computer Vision and Pattern Recognition; FOS: Electrical engineering, electronic engineering, information engineering; Electrical Engineering and Systems Science - Image and Video Processing; Machine Learning (cs.LG)

For the majority of learning-based segmentation methods, a large quantity high-quality training data is required. In this paper, we present novel model that could be trained semi- or un- supervised. Specifically, in unsupervised setting, parameterize Active contour without edges (ACWE) framework via convolutional neural network (ConvNet), and optimize parameters ConvNet using self-supervised method. another setting (semi-supervised), auxiliary ground truth used during training. We show method provides fast bone context single-photon emission computed tomography (SPECT) image.