Interactive 3-D Segmentation of the Frontal Lobe in 3.0T IR-FSPGR MR Images Using Fuzzy Rule-Based ACM
Yuji Fujiki*, Syoji Kobashi*, Mieko Matsui**, Noriko Inoue**, Katsuya Kondo*, Yutaka Hata*, and Tohru Sawada**
*Graduate School of Engineering, Himeji Institute of Technology, 2167, Shosha, Himeji, Hyogo, 671-2201, Japan
**BF Research Institute, Inc.
c/o National Cardiovascular Center, 5-7-1, Fujishirodai, Suita, Osaka, 565-0873, Japan
Received:January 30, 2003Accepted:February 2, 2003Published:June 20, 2003
Keywords:active contour model, fuzzy inference, frontal lobe, 3.0T MR image, medical image segmentation
Volumetry and surface rendering of the cerebral lobes are effective for evaluating lobar atrophy. This paper proposes a novel computer-aided system for segmenting the frontal lobe from 3-D human brain IR-FSPGR MR images with the fuzzy rule-based active contour model (ACM). The proposed system uses the criterial curves consisting of the central sulcus and the Sylvian fissure, which are given by users with the proposed graphical user interface. The user-given criterial curves are optimized by the fuzzy rule-based ACM. The fuzzy rule-based ACM can represent physicians’ knowledge with fuzzy if-then rules. With these optimized curves and the anterior and posterior commissures, the frontal lobe is segmented automatically. The experimental results on three healthy volunteers operated by an expert user and three beginner users showed that our system could segment the frontal lobe with high repeatability by any users and to any subjects.
Cite this article as:Y. Fujiki, S. Kobashi, M. Matsui, N. Inoue, K. Kondo, Y. Hata, and T. Sawada, “Interactive 3-D Segmentation of the Frontal Lobe in 3.0T IR-FSPGR MR Images Using Fuzzy Rule-Based ACM,” J. Adv. Comput. Intell. Intell. Inform., Vol.7 No.2, pp. 189-199, 2003.Data files: