Abstract Archives of the RSNA, 2008
SSM02-05
Comparison of Two Software Versions of a Commercially Available Computer-aided Detection (CAD) System for Detecting Breast Cancers
Scientific Papers
Presented on December 3, 2008
Presented as part of SSM02: Breast Imaging (Computer-aided Detection)
Soo-Yeon Kim, Presenter: Nothing to Disclose
Woo Kyung Moon MD, Abstract Co-Author: Nothing to Disclose
Nariya Cho MD, Abstract Co-Author: Nothing to Disclose
To retrospectively compare the performance of two software versions of a commercially available CAD system applied to full-field digital mammograms for detecting breast cancer in screening group
Version 8.3 and 3.1 of a CAD system (ImageChecker M1000, R2 Technology) was applied to the full-field digital mammograms (Senographe 2000D; GE Medical Systems) of 130 women (age range, 32 - 80; mean age, 52.3 years) with 130 breast cancers (93 invasive ductal (IDC), 3 lobular (ILC) carcinomas and 34 DCIS) detected by screening. Overall sensitivity and sensitivity according to lesion type and histopathology type were analyzed. CAD marks were considered positive if, on at least one view, they correctly identified the location of the corresponding mammographic lesion. To obtain an estimate of false-positive CAD mark rate, the contralateral mammograms of these women with no mammographic and clinical abnormality during 2-year follow-up were used. Differences between the data in versions 8.3 and 3.1 were analyzed by Fisher exact test.
The performance of version 8.3 and 3.1 CAD system were as follows: 96.2% (125 of 130) vs. 92.3% (120 of 130) (p < .001) in overall sensitivity, 100% (84 of 84) vs. 100% (84 of 84) in sensitivity for calcifications, 89.1% (41 of 46) vs. 78.3% (36 of 46) (p < .001) in sensitivity for masses. Version 8.3 marked five additional cancers which were missed by version 3.1. 9 cancers that were marked on both views in version 8.3 were marked on only one view in version 3.1, whereas 3 cancers that were marked on only one view in version 8.3 were marked on both views in version 3.1. Versions 8.3 and 3.1 both correctly marked all 34 DCIS and 3 ILC. However, 5 (5.3%) and 10 (10.8%) of 93 IDCs were not marked by version 8.3 and 3.1, respectively. The average false positive marks per image were 0.45 (0.13 for calcifications, 0.32 for masses) in version 8.3 and 0.38 (0.15 for calcifications, 0.23 for masses) in version 3.1.
The newer software version of the CAD system, version 8.3, applied to full-field digital mammograms showed better overall sensitivity (96.2% vs. 92.3%) for detection of breast cancer than the older one attributed to improved sensitivity for masses.
Update of the CAD software version for full-field digital mammographic images may be useful in improving the detection of invasive breast cancers which manifest only as a mass on the mammogram.
Kim, S,
Moon, W,
Cho, N,
Comparison of Two Software Versions of a Commercially Available Computer-aided Detection (CAD) System for Detecting Breast Cancers. Radiological Society of North America 2008 Scientific Assembly and Annual Meeting, February 18 - February 20, 2008 ,Chicago IL.
http://archive.rsna.org/2008/6020652.html