HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Karssemeijer, N. Articles by Holland, R. Search for Related Content PubMed PubMed Citation Articles by Karssemeijer, N. Articles by Holland, R.

Computer-aided Detection versus Independent Double Reading of Masses on Mammograms¹

¹ From the Departments of Radiology (N.K.) and Epidemiology and Biostatistics (J.D.M.O., A.L.M.V.) and the National Expert and Training Center for Breast Cancer Screening (J.H.C.L.H., R.H.), University Medical Center Nijmegen, Geert Grooteplein 18, 6525 GA Nijmegen, the Netherlands; and the Department of Public Health, National Evaluation Team for Breast Cancer Screening (J.H.G., H.J.d.K.). From the 2001 RSNA scientific assembly. Received November 30, 2001; revision requested January 10, 2002; final revision received August 5; accepted August 22. Supported by the National Evaluation Team for Breast Cancer Screening. Address correspondence to N.K. (e-mail: n.karssemeijer@rad.umcn.nl).

View larger version (20K): [in a new window]   Figure 1. Graph shows the step (solid line) and linear (dashed line) CAD weight functions used to combine the mass marker output with the observer data. The horizontal axis represents the level of normality of the CAD markers, expressed as the number of normal (ie, noncancerous) regions per image that would be hit by the CAD system on average when only the markers on regions with levels of suspicion smaller than that level of normality would be displayed.

View larger version (25K): [in a new window]   Figure 2. Graph illustrates the mean performance of 10 screening radiologists in reading 115 prior mammograms from cancer cases with a visible mass, architectural distortion, or asymmetry, and 250 mammograms from normal cases (solid line). The dashed lines show the range of performance (minimum and maximum).

View larger version (21K): [in a new window]   Figure 3. Graph shows the numbers of times the radiologists reported cancers in a subset of 50 positive cases, which collectively were determined to be cases of the most obvious cancers on prior mammograms.

View larger version (33K): [in a new window]   Figure 4. Graph illustrates mean sensitivities for the detection of visible masses on prior mammograms at single reading, independent double reading, and independent CAD reading, as functions of the false-positive fraction. Performance improved both with CAD and double reading. The improved performance with CAD was a result of improved lesion characterization and did not include the potential reduction in oversight errors.

View larger version (24K): [in a new window]   Figure 5. Graph illustrates detection performance of the CAD system in 115 study cases of visible masses on prior mammograms, as a function of the false-positive mass marker rate. For comparison, the performance of the CAD system in the same cases but with use of the diagnostic mammograms, in which the cancers were detected later, also is shown. The difference in performance between the two sets of mammograms was due to the increased visibility of the cancers on the diagnostic images and the fact that most of the prior mammograms were only mediolateral-oblique views. In the observer study, we used only the CAD results obtained from the prior mammograms.