Automatic Liver Segmentation in MR and CE-MR Images with LCVAC - GAC Approach Using Mean- shape Initialization Technique

Abstract

The Volume of the liver is a determining factor in measuring the severity of liver diseases and should be monitored regularly. Consequently, liver segmentation and volume estimation using image processing techniques play an important role in the follow up procedure. Automated liver segmentation is a challenging problem mostly addressed in CT images. MR imaging is preferred by radiologists for follow-up procedures since it does not expose patients to ionizing radiation and provides higher resolution. However, fewer studies report liver segmentation in MR images. MRI Liver segmentation represents a challenge due to presence of characteristic artifacts, such as partial volumes, noise, low contrast and poorly defined edges of the liver with respect to adjacent organs. In the present study we introduced a new automatic algorithm to 3D liver segmentation for MR and CE-MR images. The proposed algorithm includes a liver mean- shape that provides an automatic initialization together with a novel active contour method based on region with edge-weight and edge terms. In addition, different areas of liver were found, and dependent parameters were calculated automatically using modern geodesic function. We tested our algorithm on images acquired from 54 subjects in two hospitals. Finally, the results of the proposed method were compared with those of two conventional active contour methods.