Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Evaluating the performance of detection algorithms in digital mammography.

M Kallergi1, G M Carney, J Gaviria

  • 1Department of Radiology, University of South Florida, and H. Lee Moffitt Cancer Center & Research Institute, Tampa 33612, USA. kallergi@rad.usf.edu

Medical Physics
|March 17, 1999
PubMed
Summary

Evaluating computer-assisted diagnosis (CAD) algorithms for mammography requires standardized performance metrics. This study proposes new indices to accurately assess detection accuracy and types, ensuring reliable algorithm comparison.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Safety of vaginal erbium laser: A review of 113,000 patients treated in the past 8 years.

Climacteric : the journal of the International Menopause Society·2020
Same author

Improved method for automatic identification of lung regions on chest radiographs.

Academic radiology·2001
Same author

Simulation model of mammographic calcifications based on the American College of Radiology Breast Imaging Reporting and Data System, or BIRADS.

Academic radiology·1998
Same author

MRI measurement of brain tumor response: comparison of visual metric and automatic segmentation.

Magnetic resonance imaging·1998
Same author

Fragmentary window filtering for multiscale lung nodule detection: preliminary study.

Academic radiology·1998
Same author

Medical image databases for CAD applications in digital mammography: design issues.

Studies in health technology and informatics·1996

Area of Science:

  • Medical Imaging
  • Artificial Intelligence in Healthcare
  • Radiology

Background:

  • Computer-assisted diagnosis (CAD) algorithms are crucial for mammography interpretation.
  • Current evaluation methods for CAD algorithms lack standardization, leading to inconsistent performance reporting.
  • Existing performance indices offer a limited view of an algorithm's true diagnostic capabilities.

Purpose of the Study:

  • To highlight the limitations of current performance index calculations for CAD algorithms in mammography.
  • To propose novel performance indices for a more comprehensive evaluation of CAD systems.
  • To establish clear criteria for assessing detection accuracy and types in mammographic analysis.

Main Methods:

  • Analysis of various strategies for estimating true positive (TP) and false positive (FP) rates.

Related Experiment Videos

  • Development of new performance indices to account for detection degree and type (calcification, mass, artifact).
  • Testing evaluation strategies with specific criteria for area overlap, calcification clustering, and artifact detection.

    • Main Results:

    • Current TP/FP rate estimations can over- or underestimate true algorithm performance.
    • Proposed indices provide a more complete picture by considering detection completeness and classification.
    • Recommended criteria include specific thresholds for detected area, calcification cluster density, and separate handling of artifacts.

    Conclusions:

    • Standardized and detailed reporting of evaluation criteria is essential for comparing CAD methodologies.
    • New performance indices are necessary for a thorough assessment of CAD algorithm effectiveness in mammography.
    • Transparent reporting of image sets and evaluation criteria will improve understanding and advancement of CAD technology.