TY - JOUR
T1 - Fuzzy patterns and classification of functional brain images for the diagnosis of alzheimer's disease
AU - Yin, Tang Kai
AU - Chiu, Nan Tsing
PY - 2005
Y1 - 2005
N2 - Alzheimer's disease is a chronic degenerative disease of the central nervous system. Most common regional abnormalities for Alzheimer's disease are symmetric or asymmetric bilateral temporal or parietal hypoperfusion. Single-photon emission computed tomography (SPECT) is a useful tool in analyzing hypoperfusion in patients with Alzheimer's disease. The aim of this research is to provide a quantitatively automatic analysis of the SPECT scans for the diagnosis of Alzheimer's disease. A characteristic-point-based fuzzy inference classifier (CPFIC) is proposed to perform two-class classification. The closeness matrix is defined to determine the closeness between training samples, and constrained minimizations are used to systematically train the CPFIC. For comparison, experiments on nearest neighbor method and support vector machine (SVM) were also performed. In error rates, the proposed CPFIC is better than nearest neighbor method, but worse than SVM method. Although the CPFIC did not perform better than SVM in error rates, the summarizing information embedded in the patterns on characteristic points can complement SVM to provide more information to radiologists.
AB - Alzheimer's disease is a chronic degenerative disease of the central nervous system. Most common regional abnormalities for Alzheimer's disease are symmetric or asymmetric bilateral temporal or parietal hypoperfusion. Single-photon emission computed tomography (SPECT) is a useful tool in analyzing hypoperfusion in patients with Alzheimer's disease. The aim of this research is to provide a quantitatively automatic analysis of the SPECT scans for the diagnosis of Alzheimer's disease. A characteristic-point-based fuzzy inference classifier (CPFIC) is proposed to perform two-class classification. The closeness matrix is defined to determine the closeness between training samples, and constrained minimizations are used to systematically train the CPFIC. For comparison, experiments on nearest neighbor method and support vector machine (SVM) were also performed. In error rates, the proposed CPFIC is better than nearest neighbor method, but worse than SVM method. Although the CPFIC did not perform better than SVM in error rates, the summarizing information embedded in the patterns on characteristic points can complement SVM to provide more information to radiologists.
UR - http://www.scopus.com/inward/record.url?scp=23944437729&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=23944437729&partnerID=8YFLogxK
M3 - Conference article
AN - SCOPUS:23944437729
SN - 1098-7584
SP - 161
EP - 166
JO - IEEE International Conference on Fuzzy Systems
JF - IEEE International Conference on Fuzzy Systems
T2 - IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2005
Y2 - 22 May 2005 through 25 May 2005
ER -