Computer-aided diagnosis (CAD) involves a computerized analysis of rad
iographs that is used as a ''second opinion'' by the radiologist. The
approach presented incorporates computer vision and artificial intelli
gence techniques and includes schemes for the analysis of lung nodules
, interstitial infiltrates, and cardiomegaly seen on chest radiographs
; masses and clustered microcalcifications on mammograms; and stenoses
and blood flow on angiograms. The demonstration of various CAD scheme
s in chest radiography and mammography on a six-monitor workstation si
mulates one possible clinical implementation of CAD in radiology. Whet
her soft- or hard-copy display media are used, the radiologist can ref
er to the CAD results and still use the original radiograph for the fi
nal diagnosis. Although initial impressions of this simulated ''intell
igent'' workstation are encouraging, CAD is still in a preliminary sta
ge of development. Various methods for effectively and efficiently int
egrating CAD into a clinical radiology department are being investigat
ed.