Web-Based System for Remote Mammogram Processing

Abstract

This paper presents a web-based system for remote processing of mammographic images (mammograms) using Matlab Web Service. The Matlab Web Service is a toolbox for accessing Matlab applications from the internet. We created an image processing system that combines the computational and graphical abilities of Matlab with remote access through the internet. Our system for mammogram processing is a Matlab application stored on a local workstation. By using a web browser users can upload their data (mammogram and parameters necessary for image processing) and view the results obtained after processing (notification about the breast cancer existence). The presented computer system for mammogram processing includes three modules: image preprocessing, feature extraction, and classification. The preprocessing phase consists of the image de-noising, region of interest extraction and image contrast enhancement. The second phase is based on feature extraction techniques for detecting abnormalities in digital mammograms. A total of 20 texture features based on gray-level co-occurrence matrices are used to evaluate the effectiveness of textural information possessed by mass regions. To investigate the ability of the feature set in differentiating abnormal from normal tissue, we employed three different classifiers: Support vector machine classifier, Naive Bayes classifier and K-Nearest Neighbor classifier. The efficiency of classification is provided using a cross-validation technique. The main characteristics of the proposed web-based diagnostic system are that users can use application written in Matlab without installing Matlab software, and they do not need to have any knowledge about programming in Matlab to run application.