Region and Location Based Indexing and Retrieval of MR-T2 Brain Tumor Images

TL;DR

This paper presents a region and location-based indexing and retrieval system for MR-T2 brain tumor images using texture features and hash structures to improve search efficiency.

Contribution

It introduces a novel combined indexing method utilizing texture features and hash structures for efficient retrieval of brain tumor images.

Findings

Reduced search space and time through hash-based indexing.

Effective retrieval of similar brain tumor images based on texture features.

Enhanced accuracy in locating tumor regions within brain images.

Abstract

In this paper, region based and location based retrieval systems have been implemented for retrieval of MR-T2 axial 2-D brain images. This is done by extracting and characterizing the tumor portion of 2-D brain slices by use of a suitable threshold computed over the entire image. Indexing and retrieval is then performed by computing texture features based on gray-tone spatial-dependence matrix of segmented regions. A Hash structure is used to index all images. A combined index is adopted to point to all similar images in terms of the texture features. At query time, only those images that are in the same hash bucket as those of the queried image are compared for similarity, thus reducing the search space and time.