Consensus Based Medical Image Segmentation Using Semi-Supervised\n Learning And Graph Cuts

2016/12/07 by Dwarikanath Mahapatra, Mahapatra, Dwarikanath · 4 citations
Medicine · Computer Science · #Radiomics and Machine Learning in Medical Imaging #Colorectal Cancer Screening and Detection #Medical Image Segmentation Techniques

paper · pdf · doi:10.48550/arxiv.1612.02166

Abstract

Medical image segmentation requires consensus ground truth segmentations to\nbe derived from multiple expert annotations. A novel approach is proposed that\nobtains consensus segmentations from experts using graph cuts (GC) and semi\nsupervised learning (SSL). Popular approaches use iterative Expectation\nMaximization (EM) to estimate the final annotation and quantify annotator's\nperformance. Such techniques pose the risk of getting trapped in local minima.\nWe propose a self consistency (SC) score to quantify annotator consistency\nusing low level image features. SSL is used to predict missing annotations by\nconsidering global features and local image consistency. The SC score also\nserves as the penalty cost in a second order Markov random field (MRF) cost\nfunction optimized using graph cuts to derive the final consensus label. Graph\ncut obtains a global maximum without an iterative procedure. Experimental\nresults on synthetic images, real data of Crohn's disease patients and retinal\nimages show our final segmentation to be accurate and more consistent than\ncompeting methods.\n

Citations

• Inflammatory Bowel Disease Diagnosed with US, MR, Scintigraphy, and CT: Meta-analysis of Prospective Studies

Cited by

• Multimodal Generalized Zero Shot Learning for Gleason Grading using Self-Supervised Learning
• Learning of Inter-Label Geometric Relationships Using Self-Supervised Learning: Application To Gleason Grade Segmentation
• CT Image Synthesis Using Weakly Supervised Segmentation and Geometric\n Inter-Label Relations For COVID Image Analysis
• Pathological Retinal Region Segmentation From OCT Images Using Geometric Relation Based Augmentation

Related

• When Does Consensus Beat Voting? A Critical Analysis of Statistical Label Fusion in Medical Image Segmentation
• Mutual Consistency Learning for Semi-supervised Medical Image Segmentation
• Affinity-Graph-Guided Contractive Learning for Pretext-Free Medical Image Segmentation with Minimal Annotation
• MSE-Nets: Multi-annotated Semi-supervised Ensemble Networks for Improving Segmentation of Medical Image with Ambiguous Boundaries
• EAUWSeg: Eliminating annotation uncertainty in weakly-supervised medical image segmentation