Yumi Kashikura, Presenter: Nothing to Disclose

Ryohei Nakayama PhD, Abstract Co-Author: Nothing to Disclose

Akiyoshi Hizukuri, Abstract Co-Author: Nothing to Disclose

Yuki Nohara MD, Abstract Co-Author: Nothing to Disclose

Masako Yamashita MD, Abstract Co-Author: Nothing to Disclose

Tomoko Ogawa, Abstract Co-Author: Nothing to Disclose

 A computer-aided diagnosis (CAD) scheme for determining histological classifications of breast masses is expected to be useful for clinicians in the differential diagnosis. The purpose of this study was to evaluate the usefulness of the presentation of the likelihood of histological classifications evaluated by the CAD scheme in distinguishing between benign and malignant masses on ultrasonographic (US) images.

Our database consisted of 390 breast US images with masses obtained from 390 patients. Three experienced clinicians independently provided subjective ratings on the likelihood of malignancy for each of the 390 masses. Fifty benign masses (25 fibroadenomas and 25 cysts) and fifty malignant masses (25 invasive ductal carcinomas and 25 noninvasive ductal carcinomas) were selected as unknown cases for an observer study based on a stratified randomization method with the average subjective ratings. For each unknown case, the likelihood of histological classifications was evaluated by the CAD scheme using 8 objective features based on clinicians’ subjective impressions on image features that clinicians commonly use for describing masses on ultrasonographic images. During the observer study, observers provided their confidence level regarding the malignancy of the unknown case before viewing the evaluated result and after viewing the likelihood of the histological classifications. Seven clinicians participated in the observer study. The results were evaluated with multi-reader, multi-case receiver operating characteristic (ROC) analysis.

For all observers, the areas under the ROC curves (AUCs) were improved by use of the CAD scheme. The average AUC for all observers without CAD was 0.714, while that with CAD was 0.889. These differences were statistically significant (P = 0.001).

The presentation of the likelihood of histological classifications by the CAD scheme improved clinicians’ performance in the differential diagnosis of masses on US images.

The CAD scheme for determining histological classifications of masses can improve clinicians’ performance in distinguishing between benign and malignant masses on US images.

Kashikura, Y, Nakayama, R, Hizukuri, A, Nohara, Y, Yamashita, M, Ogawa, T, Improved Differential Diagnosis of Breast Masses on Ultrasonographic Images with CAD Scheme for Determining Histological Classifications.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12026495.html