Patch-Based Segmentation with Spatial Consistency: Application to MS Lesions in Brain MRI.

Roey Mechrez,Jacob Goldberger,Hayit Greenspan

doi:10.1155/2016/7952541

Abstract

This paper presents an automatic lesion segmentation method based on similarities between multichannel patches. A patch database is built using training images for which the label maps are known. For each patch in the testing image, k similar patches are retrieved from the database. The matching labels for these k patches are then combined to produce an initial segmentation map for the test case. Finally an iterative patch-based label refinement process based on the initial segmentation map is performed to ensure the spatial consistency of the detected lesions. The method was evaluated in experiments on multiple sclerosis (MS) lesion segmentation in magnetic resonance images (MRI) of the brain. An evaluation was done for each image in the MICCAI 2008 MS lesion segmentation challenge. Results are shown to compete with the state of the art in the challenge. We conclude that the proposed algorithm for segmentation of lesions provides a promising new approach for local segmentation and global detection in medical images.

Highlights

Patch-based methods have been shown to be an effective approach for labeling brain structures, as shown, for example, in [1, 2]
The segmentation is first obtained based on intensity patch similarity and further iteratively refined with the spatial label information
The proposed framework segments the multiple sclerosis (MS) image without requiring registration or an atlas. This makes the framework robust to registration errors which are likely to occur if there is a high degree of anatomical variability

Summary

Introduction

Patch-based methods have been shown to be an effective approach for labeling brain structures (and other body structures), as shown, for example, in [1, 2]. These approaches label each voxel of a target image by comparing the image patch, centered on the voxel with patches from an atlas library, and assigning the most probable label according to the closest matches. Various patch-based label fusion procedures have been proposed and were shown to produce accurate and robust segmentation. Nonlocal means denoising [6] has helped advance the field and has led to the development of various patch-based segmentation tools for medical imaging applications methods [1, 2, 7, 8]

Objectives

Results

Conclusion