Gradient directional second derivative PEH CT number correction and soft tissue preservation for Electronic cleansing in CTC images

Abstract

This dissertation proposes two electronic cleansing (EC) methods in Computed Tomography Colonography. These methods can solve the problems in EC which are the artifact partial volume effect at an AT layer, pseudo-enhancement (PEH) effect, the ambiguity between AT and ATT layers, and the artifacts at T-junctions. The first proposed EC method removes fecal-tagging material (FTM) using a simple thresholding method after applying the gradient directional second derivative (GDSD) PEH correction method to solve the PEH effect and the modified local roughness to separate AT layers from ATT layers. The second proposed EC method integrates GDSD into material fraction model to preserve PEH soft- tissue voxels and remove FTM and artifacts at AT layers and T-junctions. AT layers can be distinguished from ATT layers using the AT layer identification to preserve ATT layers. The results from clinical evaluation by the radiologist show reliability and improvement of the proposed method when comparing with the existing method and commercial software.