Analysis of Colon polyps in CT Colonography using Image Processing Techniques

Medical Image Processing (MIP) has a strong footprint in the area of pathology which has helped doctors since decades. Virtual colonoscopy or Computed Tomography Colonography (CTC) is an imaging and diagnosis procedure for colon polyp analysis based on size and shape using image processing techniques. Our research focused on the development of an automated computer-aided assessment of polyps using CTC images. The objectives of the study were a) to segment colon exactly at the mucous membrane, b) to clean the tagged fecal materials from segmented colon without loss of colonic structures, and c) measuring the smaller polyps. New image processing hybrid methods are developed with extensive preference given to domain aspects of colon polyp analysis. An improved method of boundary-based semi-automatic segmentation was developed which accurately delineated (accuracy of 94.614±0.7754% was achieved) the colon wall. A multistep technique was developed for the virtual cleansing of the colon which uses theoretical knowledge of Housenfield Units. The key findings were, the submerged colonic structures were preserved without soft tissue erosion, pseudo enhanced voxels were corrected, and the air-contrast layer was removed without losing the adjacent tissues. With the morphological image processing operator (skeletonization) and the domain aspects of deciding the polyp through height and width measurement, an automated method was developed for smaller polyp measurement. Statistically, sensitivity=87.5%, specificity=82%, accuracy=86.27%, PPV=94.45% and NPV=64.28% were achieved. The method took 1.5−2min for colon segmentation, 5