Breast Ultrasound Computer-Aided Diagnosis Approach to Improving Specificity and Decreasing Observer Variability

Abstract

The purpose of this grant is to construct an artificial neural network (ANN) to assist radiologists in differentiating benign from malignant solid breast lesions. The project was delayed due to a change in the principal investigator (PI) during this past, second year. As a result of those delays, a%cost extension for a third year has been requested and approved. Much of the work originally planned for the second year has been refocused and rescheduled for the upcoming third year. These include more data taking and some final analyses to evaluate the performance of the ANN computer models. Continuing the work from the first year, the ultrasound (US) examinations and mammograms of 100 solid breast lesions that subsequently underwent histologic confirmation were evaluated in a blinded manner. Descriptive terms were used to characterize the US and mammographic appearance of the lesions from a pre-defined lexicon. A new data collection process has been implemented for the upcoming year. Final results be deferred until the end of this year.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 1999
Accession Number
ADA366912

Entities

People

  • Joseph Y. Lo

Organizations

  • Duke University Hospital

Tags

Communities of Interest

  • Human Systems

DTIC Thesaurus Topics

  • Accuracy
  • Biomedical Research
  • Breast Cancer
  • Cancer
  • Computer-Aided Diagnosis
  • Computers
  • Consistency
  • Databases
  • Laboratory Animals
  • Materials
  • Medical Personnel
  • Neoplasms
  • Neural Networks
  • Observers
  • Physicians
  • Recombinant Dna
  • Ultrasounds

Fields of Study

  • Medicine

Readers

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

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks