Frequency-Domain Optical Mammograph

Abstract

This research project involves the analysis of a clinical data set of frequency-domain optical mammograms (^150 patients) to assess the performance of this approach to breast cancer detection. The analysis of the breast images is complemented by theoretical and experimental studies to characterize the proposed algorithms of image processing. The objective of this research is to identify the strengths and the weaknesses of the current instrument design, and to guide the design of new optical instrumentation for breast cancer detection. During the first year of this research project, we have completed the initial analysis of the optical mammograms at each one of the four wavelengths used (690, 750, 788, 856 nm). The comparison of the single- wavelength optical mammograms with the pathology reports has led to building an ROC (Receiver Operating Characteristic) curve for this method. This ROC curve shows that single wavelength optical mammograms, while potentially sensitive to breast cancer, do not perform well in terms of discrimination of benign and malignant tumors. To exploit the spectral information, we have developed a perturbation approach to analyze the four wavelength images, whose application to a subset of 19 patients has led to promising results. We are currently extending this analysis to a larger subset of data, and we are complementing it with altenative image-processing approaches.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 01, 2000
Accession Number
ADA388016

Entities

People

  • Sergio Fantini

Organizations

  • Tufts University

Tags

Communities of Interest

  • Biomedical

DTIC Thesaurus Topics

  • Breast Cancer
  • Cancer
  • Computer Science
  • Detection
  • Detectors
  • Frequency
  • Frequency Domain
  • Health Services
  • Image Processing
  • Lasers
  • Light Sources
  • Neoplasms
  • Optical Images
  • Optical Properties
  • Optics
  • Scattering
  • Two Dimensional

Fields of Study

  • Physics

Readers

  • Image Processing and Computer Vision.
  • Medical Imaging.
  • Software Engineering.

Technology Areas

  • AI & ML