A statistical approach for robust polyp detection in CT colonography

Abstract

Abstract-In this paper we describe the development of a computationally efficient computer-aided detection (CAD) algorithm based on the statistical features derived from the local colonic surface that are used for the detection of colonic polyps in computed tomography (CT) colonography. The candidate surface voxels were detected and clustered using the surface normal intersection, convexity test, region growing and Hough Transform. The main objective of this paper is the selection of the statistical features that optimally capture the convexity of the candidate surface and consequently provide a high discrimination between local surfaces defined by polyps and folds. The developed polyp detection scheme is computationally efficient (typically takes 3.9 minute per dataset) and shows 100% sensitivity for phantom polyps greater than 5mm and 87.5% sensitivity for real polyps greater than 5mm with an average of 4.05 false positives per dataset.

Development of a synthetic phantom for the selection of optimal scanning parameters in CAD–CT colonography

“…histogram and the Gaussian distribution of the Hough Points (for a detailed description of this algorithm refer to [34]). This method is able to correctly identify all polyps above 3mm but it is worth noting that this is achieved at the cost of a high level of false positives.…”

“…We have developed an automated CAD-CTC method designed to identify the colorectal polyps in CT data [34] that evaluates the local morphology of the colon wall. Initially, the colon is segmented using a seeded 3D region growing algorithm that was applied to segment the air voxels, which assures the robust identification of the colon wall.…”

“…This method is able to correctly identify all polyps above 3mm but it is worth noting that this is achieved at the cost of a high level of false positives. In order to reduce the level of false positives, statistical features [34] including the standard deviation of surface variation, ellipsoid fitting error, sphere fitting error, three axes of the ellipsoid and the Gaussian sphere radius are calculated for each candidate surface that has been identified by the convexity method described before. These features are fed into a feature normalised classifier [62] that is able to decide whether the surface under investigation belongs to a polyp or a fold.…”

“…The phantom has been specifically designed for CAD-CTC to simulate colon polyps with different shapes (pedunculated, sessile and flat) and sizes (3 to 18mm). In our studies the CT phantom data is evaluated using an automated CAD-CTC system [34] in order to determine the influence of the scanning parameters on polyp detection. A special emphasis of our study is placed on determining the minimal radiation dose that allows robust identification of colonic polyps but not at the expense of reduced sensitivity in polyp detection.…”

“…Since its introduction by Vining et al. [11] in 1994, CTC has received extensive attention from research community and many publications have emerged in areas of 3D surface rendering and visualization [15][16][17], centerline calculation [18], colon unfolding [19] and automated polyp detection [20][21][22][23][24][25][26][27][28][29][30][31][32][33][34]. Recent publications [22,23,35] indicate that the results returned by the automatic CAD-CTC polyp detection systems in the vast majority of cases closely match or even outperform the human reader performance.…”