Computer-aided detection (CAD): the case for
- S Destounis1
© BioMed Central 2004
Published: 14 July 2004
In any discussion regarding the necessity for CAD, the ultimate goals of a screening programme must be taken into consideration balanced against the cost of instituting the programme. Double reading has been shown to increase sensitivity by as much as 15%. However, the cost no matter the method is formidable. CAD, on the other hand, while initially costly, has been shown to increase the sensitivity by as much as 23%. This increase in sensitivity is reason enough to use CAD, but must be evaluated in the context of effect on positive predictive value and recall rates. Our clinic first viewed CAD with the same concerns. Our studies to date indicate no statistically significant change in positive predictive value or recall rate. Despite an independent double-blinded double read, CAD increased sensitivity by 7%. Some studies have shown that increased sensitivity with CAD is due to ductal carcinoma in situ detection, but this was not shown in our prospective study. In our group of cancers, 87% were stage 0 or stage 1 and 77% were minimal cancers (<10 mm invasive or any size ductal carcinoma in situ). However, the cancers that the double read missed and CAD prompted were all invasive with only 57% minimal disease.