Increasing the Accuracy of Mammogram Interpretation.

Abstract

The project's aim is to refine and evaluate practically a computer-based system that will provide decision support to the radiologist in interpreting mammograms and generate a standardized report of his/her findings for the referring clinician. The fully developed system will elicit from the radiologist via spoken prompts a spoken numerical scale value for each of the diagnostically most relevant perceptual features of the mammogram (usually on a 10-point scale). It will merge the values with optimal weights via a statistical prediction rule, to calculate a probability of malignancy as an advisory for the radiologist. From the pattern of feature values, the system will automatically construct a prose report of findings for the referring clinician; spoken recommendations and impressions will be accepted by the system to complete the report. The system is to be evaluated with two groups of radiologists and cases -- one from a diagnostic or referral setting, the other from a screening or community setting -- and corresponding groups of referring clinicians. Accomplishments in Year 1 have been to set criteria for study cases, locate them in hospital files, and process them into databases -- separately for the two clinical settings; develop an improved master list of perceptual features and associated response form; lay additional groundwork, conceptually and in software, for the automated report writer; advance the capability for data entry by speech recognition; and acquire system hardware.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 01, 1995
Accession Number
ADA302417

Entities

People

  • John Swets

Organizations

  • BBN Technologies

Tags

Communities of Interest

  • Biomedical

DTIC Thesaurus Topics

  • Algorithms
  • Artificial Intelligence
  • Automated Speech Recognition
  • Cancer
  • Computers
  • Databases
  • Diagnostic Imaging
  • Hard Copy
  • Health Services
  • Hospitals
  • Identification
  • Medical Personnel
  • Neoplasms
  • Physicians
  • Recognition
  • Standards
  • Surgery

Fields of Study

  • Medicine

Readers

  • Instructional Design and Training Evaluation.
  • Oncology and Biomarker-Based Cancer Detection.
  • Theoretical Analysis.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference