Abstract
A method for automatic liver tumor segmentation from computer tomography (CT) images is presented in this paper. Segmentation is an important operation before surgery planning, and automatic methods offer an alternative to laborious manual segmentation. In addition, segmentations of automatic methods are reproducible, so they can be reliably evaluated and they do not depend on the performer of the segmentation. In this work, the segmentation is performed in two stages. First a rough segmentation of tumors is obtained by simple thresholding and morphological operations. The second stage refines the rough segmentation result using fuzzy clustering and a geometric deformable model (GDM) that is fitted on the clustering result. The method was evaluated with data provided by Liver Tumor Segmentation Challenge 08, to which the method also participated. The data included 10 images from which 20 tumors were segmented. The method showed promising results.
Keywords
Source Code and Data
No source code files available for this publication.
Reviews
Xiang Deng
Friday 25 July 2008
The authors presented an automatic liver tumor segmentation technique.
In this method, the tumor is first segmented by thresholding and morphological operation, and then refined using fuzzy C-means clustering and level-set based surface smoothing.
Detailed comments:
1) A lot of empirically determined parameters are used in the "rough segmentation" stage, which could make the algorithm sensitive to data. Such effect may have been demonstrated in Tables 1-3 by the difference in performance on the training and testing datasets.
2) In Table 3, please include the evaluation table we provided as it is. You can split the results from training and testing datasets in Table 3 into two tables.