TY - GEN
T1 - High-order graph matching based feature selection for Alzheimer's disease identification
AU - Liu, Feng
AU - Suk, Heung Il
AU - Wee, Chong Yaw
AU - Chen, Huafu
AU - Shen, Dinggang
PY - 2013/10/24
Y1 - 2013/10/24
N2 - One of the main limitations of l1-norm feature selection is that it focuses on estimating the target vector for each sample individually without considering relations with other samples. However, it's believed that the geometrical relation among target vectors in the training set may provide useful information, and it would be natural to expect that the predicted vectors have similar geometric relations as the target vectors. To overcome these limitations, we formulate this as a graph-matching feature selection problem between a predicted graph and a target graph. In the predicted graph a node is represented by predicted vector that may describe regional gray matter volume or cortical thickness features, and in the target graph a node is represented by target vector that include class label and clinical scores. In particular, we devise new regularization terms in sparse representation to impose high-order graph matching between the target vectors and the predicted ones. Finally, the selected regional gray matter volume and cortical thickness features are fused in kernel space for classification. Using the ADNI dataset, we evaluate the effectiveness of the proposed method and obtain the accuracies of 92.17% and 81.57% in AD and MCI classification, respectively.
AB - One of the main limitations of l1-norm feature selection is that it focuses on estimating the target vector for each sample individually without considering relations with other samples. However, it's believed that the geometrical relation among target vectors in the training set may provide useful information, and it would be natural to expect that the predicted vectors have similar geometric relations as the target vectors. To overcome these limitations, we formulate this as a graph-matching feature selection problem between a predicted graph and a target graph. In the predicted graph a node is represented by predicted vector that may describe regional gray matter volume or cortical thickness features, and in the target graph a node is represented by target vector that include class label and clinical scores. In particular, we devise new regularization terms in sparse representation to impose high-order graph matching between the target vectors and the predicted ones. Finally, the selected regional gray matter volume and cortical thickness features are fused in kernel space for classification. Using the ADNI dataset, we evaluate the effectiveness of the proposed method and obtain the accuracies of 92.17% and 81.57% in AD and MCI classification, respectively.
UR - http://www.scopus.com/inward/record.url?scp=84885899305&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84885899305&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-40763-5_39
DO - 10.1007/978-3-642-40763-5_39
M3 - Conference contribution
C2 - 24579155
AN - SCOPUS:84897573822
SN - 9783642407628
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 311
EP - 318
BT - Medical Image Computing and Computer-Assisted Intervention, MICCAI 2013 - 16th International Conference, Proceedings
T2 - 16th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2013
Y2 - 22 September 2013 through 26 September 2013
ER -