Mark Early Baker MD, Presenter: Nothing to Disclose

Nancy A. Obuchowski PhD, Abstract Co-Author: Nothing to Disclose

Chandra Dass MD, Abstract Co-Author: Nothing to Disclose

Renee Marie Kendzierski RPh, DO, Abstract Co-Author: Nothing to Disclose

David Marc Einstein MD, Abstract Co-Author: Nothing to Disclose

Erick Marc Remer MD, Abstract Co-Author: Nothing to Disclose

Dina Finck Caroline MD, Abstract Co-Author: Nothing to Disclose

Andrew Blum MD, Abstract Co-Author: Nothing to Disclose

Luca Bogoni PhD, Abstract Co-Author: Nothing to Disclose

et al, Abstract Co-Author: Nothing to Disclose

A pilot study to assess the efficacy of computer aided detection (CAD) of colon polyps on CT colonography (CTC) when applied to initial, inexperienced radiologist interpretations.

The CTC’s on 30 patients from 7 institutions was collected; 24 with at least one polyp > 5 mm (total polyps=33) and 6 with no polyps. Using an investigational software package (ColonCAD, Siemens Medical Systems) seven primarily inexperienced readers from two institutions evaluated the CTC and marked/scored polyps using a 6 point scale before and after CAD. Results were analyzed using McNemar’s test, modified for clustered data, based upon 8 segments per patient (total=240 segments).

Without CAD, the sensitivity of polyp detection per patient was 0.583-0.833; mean=0.738 (SE=0.058). With CAD, the sensitivity of polyp detection per patient was 0.750-0.917; mean=0.845 (SE=0.041) (significantly improved at a p=0.012, 95% CI for difference [-0.183, -0.031]). Without CAD, per patient, the FP rate ranged from 0.0-0.833; mean=0.357 (SE=0.024). With CAD, per patient, the FP rate was 0.0-0.833 (SE=0.381); mean=0.381 (SE=0.025) (p value=1.00). Without CAD, the sensitivity of polyp detection per segment was 0.536-0.786; mean=0.699 (SE=0.072). With CAD, the sensitivity of polyp detection per segment was 0.714-0.893; mean=0.806 (SE=0.057) (significantly improved; p=0.018, 95% CI for difference [-0.180, -0.034]. Without CAD, per segment, the FP rate was 0.061-0.146; mean 0.090 (SE=0.015). With CAD, the FP rate per segment was 0.071-0.175; mean=0.106 (SE=0.016) (significantly more at p=0.004, 95% CI for difference [-0.026, -0.007]).

CAD for CTC significantly improves polyp detection per patient and per segment for inexperienced readers. However, CAD does increase the FP rate of polyp detection per segment.

M.E.B.: M.E.B.,L.B.: Mark Baker receives research support from Siemens Medical Systems; Luca Bogoni is employed by Siemens Medical systems

Baker, M, Obuchowski, N, Dass, C, Kendzierski, R, Einstein, D, Remer, E, Caroline, D, Blum, A, Bogoni, L, et al, , The Efficacy of Computer-aided Detection of Colorectal Polyps When Applied to Initial, Inexperienced Radiologist Interpretations of CT Colonography: A Pilot Study.  Radiological Society of North America 2005 Scientific Assembly and Annual Meeting, November 27 - December 2, 2005 ,Chicago IL. http://archive.rsna.org/2005/4415216.html