Iterative multi-path tracking for video and volume segmentation with sparse point supervision

At its core, semantic segmentation is tasked with associating pixels, or voxels, of an image with a label that corresponds to a meaningful category. As a fundamental problem in medical image computing, an impressive amount of research on the topic has been conducted in recent years, spanning methods that segment tumors in MRI volumes  (Zikic et al., 2014; Menze, 2014), airways from chest CT scans (Miyawaki et al., 2017), vessels in retinal scans (Pilch et al., 2012) or mitochondria in electron microscopes (Seyedhosseini et al., 2013) to name a few.
Source: Medical Image Analysis - Category: Radiology Authors: Source Type: research