Deep learning and information fusion based classification of breast cancer in mammography images

Student thesis: Doctoral Thesis

Abstract

Breast cancer is one of the most dangerous diseases that can afflict especially women. Computer-aided diagnosis (CADx/CAD) systems may help radiologists in fast and accurate decision making to detect early stage of breast cancer and reduce mortality. By fusion of different techniques in image processing, machine learning (ML), and deep learning (DL), several techniques are proposed in this thesis for analysing breast cancer images, aiming to classify normal/ abnormal lesions in mammography images. The classification method is generally implemented using ML classifiers and DL methods based on the deep convolutionalneural networks (DCNN). This thesis presents new methods, assembled on recently developed DL models based on different fusion techniques to develop three different frameworks. The first framework presents an approach for segmenting the region of interest (ROI) followed by classifying it using DCNN. In the second framework, a fusion-based novel approach is presented to classify the mammogram lesions using several DCNNs. Additionally, it employs some ML techniques to improve the classification accuracy. However, the third framework introduces the decision level fusion rather than feature fusion forming a second stage classification to improve the classification accuracy as well. Overall, the methods proposed in this thesis achieved promising classification accuracy resultsthat improve the performance of the state-of-the-art approaches and may help to improve the diagnosis of breast cancer. The methodologies presented in this work are evaluated on several publicly available datasets, including the digital database for screening mammography (DDSM), the curated breast imaging subset of DDSM (CBIS-DDSM), and the mammographic image analysis society digital mammogram dataset (MIAS). Considering their limitations, a new mammogram dataset, “DAR-Breast,” is collected from the Armed Forces Hospital in Egypt, the first such in Egypt, to benefit the advancement in this area.
Date of Award20 Dec 2022
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SupervisorStephen Marshall (Supervisor) & Jinchang Ren (Supervisor)

Cite this

'