A Novel Pattern Classification Technique for the Diagnosis of Breast Cancer


Brijesh Verma;
Post Doctoral Research Fellow: Rinku Panchal.

Student: Peter McLeod, Syed Zahid Hassan, Hong Lee.

Abdesselam Bouzerdoum;
Arcot Sowmya;
Stuart Crozier;
Svetha Venkatesh.

Introduction: This project aims to study suspicious areas and their classification in digital mammograms for the development of a fast and reliable computer aided diagnostic system for breast cancer screening.
Significance: The system will be capable of recalling previously seen patterns and classifying new patterns with high accuracy.
Challenges: The main aims of this project are as follows. 1. Investigate a new feature extraction technique based on soft-clustering and 3D information from suspicious areas in digital mammograms. 2. Investigate a novel algorithm based on accuracy and genetic algorithms to find the significant of features extracted from suspicious areas in digital mammograms. The aim is to find a feature or features with consistent classification accuracy. 3. Investigate a novel learning algorithm for the classification of features into benign and malignant classes. The aim is to achieve 100% accuracy rate on seen/training mammograms (memorisation/association ability), high accuracy rate (95-100%) on unseen/test mammograms (generalisation ability) and fast training of the network.
top of page