Comparative Performance of Two Polyp Detection Systems on CT Colonography

Abstract

One polyp detection system tended to operate with a higher sensitivity, whereas the other tended to operate with a lower false-positive rate. Prospective trials using polyp detection systems as a primary or secondary means of CT colonography interpretation appear warranted.

Computer-aided Detection in Computed Tomography Colonography with Full Fecal Tagging: Comparison of Standalone Performance of 3 Automated Polyp Detection Systems

“…Regarding the sensitivity:FP ratio for lesions !10 mm in size, however, CAD system 3 showed the most favorable result, with a rather low rate of FP findings. Interestingly, system 3 showed, in our patient cohort, a higher sensitivity for the detection of lesions !10 mm in size compared with the sensitivity level reported by Fletcher et al [11] for the same system.…”

“…The effect of intravenous contrast administration and fecal tagging on the performance of the CTC-CAD systems remains unclear. However, with regard to the results of system 3 and in comparison with the previous comparative study [11], the effect of fecal tagging on CAD sensitivity cannot be considered negative. Further limitations are the approximate evaluation of causes for FP findings after CAD analysis and the rather small patient collective in our study.…”

“…Beside technical factors, readers' expertise, fatigue, distraction, and perceptual errors are considered important factors that affect polyp detection by CTC. Perceptual errors may result from the misinterpretation of normal structures, such as colonic folds, or variable conspicuity of polypoid structures [11]. Although the interpretation of CTC studies that primarily used 3D endoluminal views may increase the individual reader's performance, it can also contribute to perceptual errors because of the necessity of a bidirectional data analysis.…”

“…For different CAD systems and different patient cohorts, reported sensitivities for the standalone performance of CAD ranged between 80% and 90%, with FP rates that ranged between 2.1 and 32 FP findings per patient [14,17e24]. In a recent comparative study, Fletcher et al [11] investigated CAD on 65 CTC data sets when using 2 different CAD systems. Analysis with the first CAD system yielded 96% sensitivity for lesions !10 mm, with an average FP rate of 5.2 per patient (range, 1-26).…”

“…Computer-aided detection (CAD) systems were developed and are used as adjuncts for digital mammography and detection of pulmonary nodules on chest CT (computed tomography) [9,10]. A recent study on CAD in CTC compared 2 systems on data without fecal tagging [11]. However, both stool and fluid tagging are considered routine techniques in CTC at most clinical institutions.…”