Min Woo Lee MD, Presenter: Nothing to Disclose

Se Hyung Kim, Abstract Co-Author: Nothing to Disclose

Hee Sun Park MD, Abstract Co-Author: Nothing to Disclose

June-Goo Lee, Abstract Co-Author: Nothing to Disclose

Jeong Min Lee, Abstract Co-Author: Nothing to Disclose

Jae Young Lee MD, Abstract Co-Author: Nothing to Disclose

Joon Koo Han MD, Abstract Co-Author: Nothing to Disclose

Byung Ihn Choi MD, Abstract Co-Author: Nothing to Disclose

et al, Abstract Co-Author: Nothing to Disclose

et al, Abstract Co-Author: Nothing to Disclose

To compare the diagnostic performances of commercial and academic CAD systems for polyp detection on CTC using pig phantoms.

103 polyps (48 <6mm, 55 ≥6mm) in 13 phantoms were analyzed. There were 33 flat, 45 sessile, and 25 pedunculated polyps. After CTC was performed using a 64-row MDCT, two commercial (CAD A and B) and one academic CAD systems were applied to each CTC dataset. Per-polyp sensitivities regarding the size and shape were calculated and compared for each CAD system. The mean number and causes of false-positives (FPs) were analyzed.

Overall per-polyp sensitivity for all polyps was 64.1% for both CAD A and B, and 83.5% for academic CAD (p0.05). According to morphology, per-polyp sensitivities were 51.5%, 57.6%, 81.8% for flat polyps; 60.0%, 62.2%, and 84.4% for sessile; and 88.0%, 76.0%, and 62.5% for pedunculated, for CAD A, B, and academic CAD. The mean number of FPs was 3.5 for CAD A, 2.6 for CAD B, and 4.6 for academic CAD (P>0.05). The most common cause of FPs was extracolonic structure (41.2% in academic CAD), followed by balloon catheter (22.4%, 50.0%, and 40.0%, respectively), and prominent folds (34.7% in CAD A) (P<0.001).

The performance of the three CAD systems was comparable for polyps ≥6mm. In addition, the number of FPs was not significantly different among the CAD systems. However, the distribution of the causes of FPs was significantly different among the systems.

Different CAD systems show comparable performance but different distribution of FPs. Radiologists should recognize the effect of different algorithms on CTC CAD characteristics.

Lee, M, Kim, S, Park, H, Lee, J, Lee, J, Lee, J, Han, J, Choi, B, et al, , et al, , An Anthropomorphic Phantom Study of Computer-aided Detection Performance for Polyp Detection on CT Colonography: Comparison of Commercially and Academically Available Systems.  Radiological Society of North America 2007 Scientific Assembly and Annual Meeting, November 25 - November 30, 2007 ,Chicago IL. http://archive.rsna.org/2007/5008859.html