HDF: Heat diffusion fields for polyp detection in CT colonography

Computer aided detection (CAD) in computed tomography colonography (CTC) aims at detecting colonic polyps that are the precursors of cancer. We propose to use heat diffusion patterns for polyp detection and identification in CTC. The underlying idea of the proposed method is to utilize the heat diffusion process, which is closely related to the diffusion filtering, to generate a vector field (HDF) that is correlated with the shape of the colon wall. The HDF and a local voxel distribution parameter were used for detection and identification. We evaluated the method on real patient CTC data acquired from eight volunteers. 100% sensitivity is achieved at 0.53 false positives per patient for polyps >=8mm and 78% sensitivity is achieved at 0.53 false positives per patient when all polyps >=5mm are considered.