TY - GEN
T1 - GA and AdaBoost-based feature selection and combination for automated identification of dementia using FDG-PET imaging
AU - Xia, Yong
AU - Zhang, Zhe
AU - Wen, Lingfeng
AU - Dong, Pei
AU - Feng, David Dagan
PY - 2012
Y1 - 2012
N2 - FDG-PET imaging offers the potential for an image-based automated identification of different dementia syndromes. However, various global and local FDG-PET image features have their limitations in characterizing the patterns of this disease. In this paper, we propose an automated approach to identifying the patients with suspected Alzheimer's disease, patients with frontotemporal dementia and normal controls based on the jointly using a group of global features and three groups of local features extracted from parametric FDG-PET images. In this approach, we employ the genetic algorithm to select the features that have best discriminatory ability, and use the AdaBoost technique to adaptively combine four feature groups in constructing a strong classifier. We compared our approach to other classification methods in 154 clinical FDG-PET studies. Our results show that, with the complementary use of the selected global and local features, the proposed approach can substantially improve the accuracy of FDG-PET imaging-based dementia identification.
AB - FDG-PET imaging offers the potential for an image-based automated identification of different dementia syndromes. However, various global and local FDG-PET image features have their limitations in characterizing the patterns of this disease. In this paper, we propose an automated approach to identifying the patients with suspected Alzheimer's disease, patients with frontotemporal dementia and normal controls based on the jointly using a group of global features and three groups of local features extracted from parametric FDG-PET images. In this approach, we employ the genetic algorithm to select the features that have best discriminatory ability, and use the AdaBoost technique to adaptively combine four feature groups in constructing a strong classifier. We compared our approach to other classification methods in 154 clinical FDG-PET studies. Our results show that, with the complementary use of the selected global and local features, the proposed approach can substantially improve the accuracy of FDG-PET imaging-based dementia identification.
KW - AdaBoost algorithm
KW - dementia classification
KW - FDG-PET imaging
KW - genetic algorithm
KW - support vector machine
UR - https://www.scopus.com/pages/publications/84865812182
U2 - 10.1007/978-3-642-31919-8_17
DO - 10.1007/978-3-642-31919-8_17
M3 - 会议稿件
AN - SCOPUS:84865812182
SN - 9783642319181
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 128
EP - 135
BT - Intelligent Science and Intelligent Data Engineering - Second Sino-Foreign-Interchange Workshop, IScIDE 2011, Revised Selected Papers
T2 - 2nd Sino-Foreign-Interchange Workshop on Intelligent Science and Intelligent Data Engineering, IScIDE 2011
Y2 - 23 October 2011 through 25 October 2011
ER -