Co-occurrence analysis for discovery of novel breast cancer pathology patterns

IEEE Trans Inf Technol Biomed. 2006 Jul;10(3):497-503. doi: 10.1109/titb.2005.863863.

Abstract

To discover novel patterns in pathology co-occurrence, we have developed algorithms to analyze and visualize pathology co-occurrence. With access to a database of pathology reports, collected under a single protocol and reviewed by a single pathologist, we can conduct an analysis greater in its scope than previous studies looking at breast pathology co-occurrence. Because this data set is unique, specialized methods for pathology co-occurrence analysis and visualization are developed. Primary analysis is through a co-occurrence score based on the Jaccard coefficient. Density maps are used to visualize global co-occurrence. When our co-occurrence analysis is applied to a population stratified by menopausal status, we can successfully identify statistically significant differences in pathology co-occurrence patterns between premenopausal and postmenopausal women. Genomic and proteomic experiments are planned to discover biological mechanisms that may underpin differences seen in pathology patterns between populations.

Publication types

  • Research Support, Non-U.S. Gov't
  • Research Support, U.S. Gov't, Non-P.H.S.

MeSH terms

  • Algorithms
  • Artificial Intelligence*
  • Biopsy / methods*
  • Breast Neoplasms / classification
  • Breast Neoplasms / pathology*
  • Cluster Analysis
  • Female
  • Humans
  • Image Enhancement / methods*
  • Image Interpretation, Computer-Assisted / methods*
  • Information Storage and Retrieval / methods*
  • Pattern Recognition, Automated / methods*
  • Reproducibility of Results
  • Retrospective Studies
  • Sensitivity and Specificity
  • User-Computer Interface