Compression of 4D medical image and spatial segmentation using deformable models

Abstract

Medical imaging has become an essential tool to assist physicians in clinical routines. The problem of huge data volume for storage or communications becomes more acute for 4D medical images (i.e., 3D volumetric images over time). A new 3D motion compensated lossy-to-lossless 4D medical image compression scheme using the integer wavelet transform is proposed. Another major issue in medical imaging is automatic or semi-automatic extraction of critical diagnostic information. In our proposed minimal path deformable model based approach, the initialization for the segmentation process is significantly simplified into a single starting point and precise and robust segmentation is achieved. Extracting complex vasculature from 3D magnetic resonance angiography is also studied. Inspired by the capillary action, a capillary active contour is modeled to extract blood vessels from MRA images. Experimental results show that the capillary active contour can achieve more precise segmentation when compared with other state-of-the-art algorithms.