Scale-Space and Edge Detection Using Anisotropic Diffusion
IEEE Transactions on Pattern Analysis and Machine Intelligence
Digital Image Processing
Hi-index | 0.00 |
A semi-automated method for segmentation of liver nodules in Computed Tomography studies is described in this paper. The application is part of a liver cancer computer-aided diagnosis (CAD) system. Its main body consists of the three-dimensional anisotropic diffusion filtering and the adaptive region growing, supported by the fuzzy inference system. Such a workflow enables elimination of noise within the image data, enhances nodule region boundaries, and cuts ,,segmentation leaks". The outcome is interactively presented to the physician with a possibility left to make manual adjustments. The system has been evaluated using a database of 17 abdominal Computed Tomography studies including 30 various liver nodules outlined by the radiologist, yielding 77% effectiveness (23 cases).