Radiology -- eLetters for Yang et al., 244 (1) 104-111

HOME

HELP

FEEDBACK

SUBSCRIPTIONS

ARCHIVE

QUICK SEARCH:

[advanced]

	Author:		Keyword(s):
Year:		Vol:		Page:

Electronic Letters to:

Breast Imaging:
Sang Kyu Yang, Woo Kyung Moon, Nariya Cho, Jeong Seon Park, Joo Hee Cha, Sun Mi Kim, Seung Ja Kim, and Jung-Gi Im
Screening Mammography–detected Cancers: Sensitivity of a Computer-aided Detection System Applied to Full-Field Digital Mammograms
Radiology 2007; 244: 104-111 [Abstract] [Full text] [PDF]

eLetters: Submit a response to this article

Electronic letters published:

Overestimating CAD Benefit in Mammography: Stefano Ciatto, MD (16 July 2007)

Overestimating CAD Benefit in Mammography 16 July 2007

Stefano Ciatto, MD,
radiologist
Centro per lo Studio e la Prevenzione Oncologica
Send letter to journal:
Re: Overestimating CAD Benefit in Mammography

s.ciatto{at}cspo.it Stefano Ciatto, MD

Editor:
I do not think that the article by Dr Yang and colleagues (1) adds evidence in favor of or against CAD use in reading mammograms. Testing CAD retrospectively on screening-detected cancers is not the proper measure: These cancers have already been detected with standard mammography, and CAD cannot improve their detection. If CAD has any use, it is to reduce the interval cancer rate, and it is on screening mammograms followed by interval cancers that is should be tested retrospectively. In fact, CAD sensitivity for interval cancers is much lower than that for screening-detected cancers (2). Moreover, the fact that CAD marks a cancer does not necessarily mean that CAD will improve sensitivity compared with standard mammography. The benefit of CAD occurs when CAD marks allow detection of cancers that would otherwise have been overlooked. With the enormous excess of false-positive results with CAD (1.8 false-positives per case in a screening setting, with a cancer detection rate usually below 1%, means that the positive predictive value of CAD—per case—is at most 1%), radiologists may tend to discount CAD markings, and it may happen that cancers marked by CAD are ruled out as benign. On the other hand, the alternative of referring for diagnostic assessment all patients with CAD marks (approximately 100%) is unacceptable. The benefits of CAD can be estimated only by prospective analysis of cancers detected or of the interval cancer rate in comparable cohorts undergoing standard and CAD reading.
References
1. Yang SK, Moon WK, Cho N, et al. Screening mammography–detected cancers: sensitivity of a computer-aided detection system applied to full-field digital mammograms. Radiology 2007;244:104-111.
2. Ciatto S, Rosselli Del Turco M, Burke P, Visioli C, Paci E, Zappa M. Comparison of standard and double reading and computer aided detection (CAD) of interval cancers at prior negative screening mammograms blind review. BJC 2003;89:1645-1649.