Computerized localization of breast lesions from two views - An experimental comparison of two methods

RATIONALE AND OBJECTIVES. The authors compared two computerized methods, th e are and cartesian straight- line, for the localization of breast lesions in two mammographic views. METHODS. A total Of 571 craniocaudal and 571 mediolateral oblique matched m ammographic image pairs (or 1142 individual images) depicting 290 pathology -verified masses on both views were selected from our image database. Using a previously developed computer-aided detection scheme, all 290 masses and 3992 suspicious but negative regions were identified. After pairing all id entified regions from both views, all masses (true-positive-true-positive m atched pairs) and a total of 10330 false-positive pairs (including false-po sitive-false-positive, true-positive-false-positive, and false-positive-tru e positive pairs) were assessed as to their position in relation to the nip ple using both the are and the cartesian straight-line methods. Receiver op erating characteristic methodology was used to evaluate the performance lev els for each method in determining, based solely on location, whether a pai r of suspicious regions represented a true mass or a false-positive combina tion, RESULTS. The areas under the receiver operating characteristic curves (A(z) ) were 0.79 and 0.78 for the arc and cartesian straight-line methods, respe ctively. The difference between the two techniques las measured by A(z)) wa s not statistically significant (P > 0.99). CONCLUSIONS. These preliminary results demonstrated that the two methods ar e comparable in identifying true masses from triangulated observations on t wo views, However, the are method is somewhat favorable because only the ni pple location is required for localization.