Spectral Analysis of Breast Cancer on Tissue Microarrays: Seeing Beyond Morphology

Abstract

Spectral imaging is a mechanism of tissue examination that combines spectral (color) with spatial information. The Varispec(exp TM) device fractionates the light into 10 nm bands and then collects images containing a frill spectral profile for each pixel. The resulting spectral profile is then analyzed using a genetic algorithm based software package called GENIE. GENIE allows combination of spectral and spatial information for binary inclusion into user defined classes. This IDEA grant proposed examination of the spectral signatures of breast cancer to distinguish benign cells from malignant cells allowing diagnosis, classification, and possibly prediction of outcomes in breast cancer. The first stages of the grant analyzed tissue microarray spots, to assess our ability to distinguish benign from malignant tissue using very "easy" cases. This has been completed with good sensitivity and specificity and more challenging specimens are now being assessed. This has progressed to more challenging cases, but the technology failed to distinguish outcome. The second part of the proposal focused on application of this technology to cytology specimens. Due to some difficulties in obtaining optimal material for breast cytology, urine cytology specimens have been used as a model system. Using this model, we have been able to distinguish benign from malignant cells with high accuracy and even shown good accuracy in adjudication of "atypical" specimens.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 01, 2005
Accession Number
ADA436898

Entities

People

  • David L Rimm

Organizations

  • Yale University

Tags

Communities of Interest

  • Air Platforms
  • Engineered Resilient Systems
  • Space

DTIC Thesaurus Topics

  • Accuracy
  • Algorithms
  • Breast Cancer
  • Cancer
  • Detection
  • Epithelial Cells
  • Genetic Algorithms
  • Health Services
  • Image Processing
  • Information Science
  • Machine Learning
  • Materials
  • Medical Personnel
  • Neoplasms
  • Pattern Recognition
  • Remote Sensing
  • Supervised Machine Learning

Readers

  • Image Processing and Computer Vision.
  • Molecular Biology and Genetics
  • Molecular and genetic basis of cancer.

Technology Areas

  • AI & ML
  • Biotechnology