Back to search

South Dakota State University

A New Approach for Segmentation of Breast MR Images and Object-retrieval Based on Correlation Between Different Types of Breast Images

Abstract

dc:description.abstract

Magnetic Resonance Imaging (MRI) has been used as the significant mean of diagnosing breast cancers. However, due to its high cost, several alternatives have been proposed recently. The motivation of the new approach in this paper is how to replace MRI with the alternatives which do not have enough diagnosis samples so that the reliability is not sufficient. In order to raise the reliability of the alternatives, we need to find correlations between MR images and the alternatives so that the enormous samples of MRI could be used to diagnose breast cancers with the more affordable methods. This paper introduces two main issues; how to segment the fuzzy medical images and how to find correlations between two different types of breast images and retrieve the most similar objects from them based on the correlations. The goal of the new segmentation approach is to achieve simplicity, generality, and constancy which are important to both accuracy and feasibility of object extraction and analysis. The tumor objects extracted from the segmentation method will be used to evaluate the object retrieval. The major goal of the proposed retrieval method is to find the most similar objects to the query object from different types of images based on the found correlation considering all possible classes such as intensity, number of pixels, and so on. The segmentation method was simulated with 20 MRI images of different characteristics and the object retrieval was experimented with the 10 extracted objects of tumors from the MR images and 10 made-out objects. The performance of accuracy will be represented in the result and discussion section.

Degree

thesis:*
Name thesis:degree_name
Master of Science (MS)
Level thesis:degree_level
Thesis - Open Access
Discipline thesis:degree_discipline
Electrical Engineering and Computer Science

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Kang, Donghoon
Contributors dc:contributor
  • Sung Shin

Rights

dc:rights

Identifiers

dc:identifier.*
Repository record dc:identifier
https://openprairie.sdstate.edu/etd2/1941

Chain of custody

source
Harvested from
South Dakota State University
Base URL
openprairie.sdstate.edu/do/oai/
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Kang, Donghoon. A New Approach for Segmentation of Breast MR Images and Object-retrieval Based on Correlation Between Different Types of Breast Images. Thesis - Open Access thesis, https://openprairie.sdstate.edu/etd2/1941