Computerized Identification of Normal Mammograms

Abstract

The purpose of this concept award project is to develop an automated method to identify normal mammograms, that is those without breast disease. This is a new paradigm in computer-aided diagnosis (CAD), since all other CAD schemes identify breast cancer. We are relying on the natural pattern of glandular tissue in the normal breast, which radiates out from the nipple. Breast cancer disturbs this pattern. We have developed a database or 300 regions of interest (ROIs) of normal breast tissue and 200 regions containing a portion of a breast cancer. Each region was automatically extracted from a mammogram that was reduced in size and preprocessed using a wavelet filter. We are using these ROIs to train an artificial neural network called a self-organizing map (SOM) to learn the mammographic pattern of normal breast tissue. SOM are self-learning classifiers that categorize input data into a use-defined number of distinct classes. To date, we have been unsuccessful in training the SOM to categorize normal and abnormal ROIs in a reliable manner. We are in the process of changing our training protocol.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 01, 2004
Accession Number
ADA433850

Entities

People

  • Robert Nishikawa

Organizations

  • University of Chicago

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Artificial Intelligence Software
  • Biomedical Research
  • Breast Cancer
  • Cancer
  • Computers
  • Data Sets
  • Databases
  • Detection
  • Machine Learning
  • Neoplasms
  • Neural Networks
  • Supervised Machine Learning
  • Training
  • Two Dimensional
  • Unsupervised Machine Learning

Fields of Study

  • Medicine
  • Physics

Readers

  • Neural Network Machine Learning.
  • Oncology and Biomarker-Based Cancer Detection.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks