Two Methods for ICH Segmentation

This paper presents and compares two methods for automatic segmentation of computed tomography (CT) head images of human spontaneous intracerebral brain hemorrhage (ICH). Both methods have a hierarchical structure with two-level: the higher and the lower level. The first method is based on unsupervised fuzzy C-means (UFCM) and image labeling algorithm. The second method is based on UFCM and rule-based systems. The methods segment the input CT image at the higher level into a number of spatially localized regions having uniform brightness. The UFCM algorithm is used to break the input CT images. A label from the predefined label set is assigned to each of the regions obtained by UFCM. The first method performs labeling using backtracking tree search as an image labeling algorithm while the second method uses a rule-based system. The label set is composed of these labels: background, skull, brain, ICH, and calcifications. The lower segmentation level uses results of the higher segmentation level to perform further segmentation of the brain region and localization of fine structures in the brain region such as edema region. The UFCM algorithm is used by the first method to segment brain region into the edema region, ventricle, and the rest of the brain. The second method uses rule-based system to refine the ICH region and to localize the edema region. The methods have been compared and tested on real CT head images.

[1]  A. Dhawan,et al.  Knowledge-based analysis and understanding of medical images. , 1990, Computer methods and programs in biomedicine.

[2]  Isak Gath,et al.  Unsupervised Optimal Fuzzy Clustering , 1989, IEEE Trans. Pattern Anal. Mach. Intell..

[3]  R A Peters,et al.  Automatic segmentation of ultrasound images using morphological operators. , 1991, IEEE transactions on medical imaging.

[4]  James C. Bezdek,et al.  A comparison of neural network and fuzzy clustering techniques in segmenting magnetic resonance images of the brain , 1992, IEEE Trans. Neural Networks.

[5]  L O Hall,et al.  Review of MR image segmentation techniques using pattern recognition. , 1993, Medical physics.

[6]  J. Udupa,et al.  Shape-based interpolation of multidimensional objects. , 1990, IEEE transactions on medical imaging.

[7]  Jenq-Neng Hwang,et al.  A neural network-based stochastic active contour model (NNS-SNAKE) for contour finding of distinct features , 1995, IEEE Trans. Image Process..

[8]  Lawrence O. Hall,et al.  Knowledge-based classification and tissue labeling of magnetic resonance images of human brain , 1993, Electronic Imaging.

[9]  Linda G. Shapiro,et al.  Computer and Robot Vision , 1991 .

[10]  Sven Loncaric,et al.  Hierarchical segmentation of CT head images , 1996, Proceedings of 18th Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[11]  Benoit M. Dawant,et al.  Neural-network-based segmentation of multi-modal medical images: a comparative and prospective study , 1993, IEEE Trans. Medical Imaging.