Who could benefit the most from using a computer-aided detection system in full-field digital mammography?

Abstract

BackgroundThe computer-aided detection (CAD) system on mammography has the potential to assist radiologists in breast cancer screening. The purpose of this study is to evaluate the diagnostic performance of the CAD system in full-field digital mammography for detecting breast cancer when used by dedicated breast radiologist (BR) and radiology resident (RR), and to reveal who could benefit the most from a CAD application.MethodsWe retrospectively chose 100 image sets from mammographies performed with CAD between June 2008 and June 2010. Thirty masses (15 benign and 15 malignant), 30 microcalcifications (15 benign and 15 malignant), and 40 normal mammography images were included. The participating radiologists consisted of 7 BRs and 13 RRs. We calculated the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for total, normal plus microcalcification and normal plus mass both with and without CAD use for each reader. We compared the diagnostic performance values obtained with and without CAD use for the BR and RR groups, respectively. The reading time reviewing one set of 100 images and time reduction with CAD use for the BR and RR groups were also evaluated.ResultsThe diagnostic performance was generally higher in the BR group than in the RR group. Sensitivity improved with CAD use in the BR and RR groups (from 81.10 to 84.29% for BR; 75.38 to 77.95% for RR). A tendency for improvement in all diagnostic performance values was observed in the BR group, whereas in the RR group, sensitivity improved but specificity, PPV, and NPV did not. None of the diagnostic performance parameters were significantly different. The mean reading time was shortened with CAD use in both the BR and RR groups (111.6 minutes to 94.3 minutes for BR; 135.5 minutes to 109.8 minutes for RR). The mean time reduction was higher for the RR than that in the BR group.ConclusionsCAD was helpful for dedicated BRs to improve their diagnostic performance and for RRs to improve the sensitivity in a screening setting. CAD could be essential for radiologists by decreasing reading time without decreasing diagnostic performance.

Artificial Intelligence-Based Thyroid Nodule Classification Using Information from Spatial and Frequency Domains

“…Consequently, the performance of diagnosis is limited and varies with the doctor’s experience. To surmount this limitation, a double screening scheme has been applied in some hospitals by employing an additional expert [1]. However, this scheme is expensive and time consuming.…”

Does Breast Imaging Experience During Residency Translate Into Improved Initial Performance in Digital Breast Tomosynthesis?

“…Because of the increased interpretation time for DBT, the total number of cases interpreted by each trainee (n ¼ 60) was smaller than the number in some similar mammography studies (range: 50-150). However, we included more readers (n ¼ 29) than are involved in comparable mammography studies (range: [10][11][12][13][14][15][16][17][18][19], to compensate for this difference [11,12,[14][15][16]. Thus, we were able to evaluate 1,740 data points (60 studies, times 29 readers), which provides similar, if not better, statistical power than prior work evaluating resident performance in mammography.…”

Usefulness of grayscale inverted images in addition to standard images in digital mammography

“…In the study by Fiona et al, the double-reading system and CAD system had similar detection rates [ 6 ]. However, another study suggested that the CAD system improved diagnostic performance and reduced assessment time [ 14 ]. An additional benefit of the CAD system is its capacity to detect microcalcification even when used by radiologists with little experience [ 15 , 16 ].…”