Feature Selection Framework of Whole-Brain Functional Magnetic Resonance Imaging Data Based on Regularized Softmax Regression
QU Yongkang1, JI Junzhong1, LIANG Peipeng2, GAO Mingxia1
1.Beijing Municipal Key Laboratory of Multimedia and Intelligent Software Technology, College of Computer Science, Beijing University of Technology, Beijing 100124 2.Xuanwu Hospital, Capital Medical University, Beijing 100053
Abstract To solve the classification model overfitting problem caused by the high dimension and small sample properties of functional magnetic resonance imaging (fMRI) data, a feature selection framework of whole-brain fMRI data combining L1-norm regularization and L2-norm regularization in softmax regression is proposed. Firstly, the whole brain is divided into the region of interest (ROI) and the region of non-interest (RONI) in terms of the characteristics of brain cognition. Then, L2-norm regularization shrinking the weighting coefficients is used to model all voxels in ROI while L1-norm regularization with a sparse effect is employed for modeling the activated voxels in RONI. Finally, the regularized softmax regression model of whole-brain fMRI data is constructed by integrating all voxels in ROI and the activated voxels in RONI. The experimental results on Haxby datasets show that the regularization strategies of L2-norm and L1-norm effectively improve the whole-brain classification performance compared to some other methods.
About author:: QU Yongkang, born in 1989, master student. His research interests include machine learning and biological data miningJI Junzhong(Corresponding author), born in 1969, Ph.D., professor. His research interests include machine learning, data mining, swarm intelligence and bioinformatics.LIANG Peipeng, born in 1979, Ph.D., associate professor. His research interests include medical imaging and cognitive neuroscience.GAO Mingxia, born in 1973, Ph.D.. Her research interests include data mining, semantic web and knowledge engineering.
QU Yongkang,JI Junzhong,LIANG Peipeng等. Feature Selection Framework of Whole-Brain Functional Magnetic Resonance Imaging Data Based on Regularized Softmax Regression[J]. , 2016, 29(7): 641-649.
[1] 李恩中.功能磁共振成像在认知神经科学中的应用.计算机科学与探索, 2008, 2(6): 589-600. (LI E Z. The Application of BOLD-fMRI in Cognitive Neuroscience. Journal of Frontiers of Computer Science and Technology, 2008, 2(6): 589-600.) [2] PEREIRA F, MITCHELL T, BOTVINICK M. Machine Learning Classifiers and fMRI: A Tutorial Overview. NeuroImage, 2009, 45(1): S199-S209. [3] LEMM S, BLANKERTZ B, DICKHAUS T, et al. Introduction to Machine Learning for Brain Imaging. NeuroImage, 2011, 56(2): 387-399. [4] 邓红霞,相 洁,游 雅,等.基于fMRI的思维数据分析方法研究.计算机研究与发展, 2015, 51(4): 773-780. (DENG H X, XIANG J, YOU Y, et al. Analysis Method of Thinking Data Based on fMRI. Journal of Computer Research and Development, 2015, 51(4): 773-780.) [5] 刘建伟,崔立鹏,刘泽宇,等.正则化稀疏模型.计算机学报, 2015, 38(7): 1307-1325. (LIU J W, CUI L P, LIU Z Y, et al. Survey on the Regularized Sparse Models. Chinese Journal of Computers, 2015, 38(7): 1307-1325.) [6] MWANGI B, TIAN T S, SOARES J C. A Review of Feature Reduction Techniques in Neuroimaging. Neuroinformatics, 2014, 12(2): 229-244. [7] CHOU C A, KAMPA K B, MEHTA S H, et al. Information-Theoretic Based Feature Selection for Multi-voxel Pattern Analysis of fMRI Data // Proc of the International Conference on Brain Informatics. Berlin, Germany: Spring-Verlag, 2012: 196-208. [8] DE MARTINO F, VALENTE G, STAEREN N, et al. Combining Multivariate Voxel Selection and Support Vector Machines for Mapping and Classification of fMRI Spatial Patterns. NeuroImage, 2008, 43(1): 44-58. [9] GUYON I, ELISSEEFF A. An Introduction to Variable and Feature Selection. Journal of Machine Learning Research, 2003, 3: 1157-1182. [10] GUYON I, WESTON J, BARNHILL S, et al. Gene Selection for Cancer Classification Using Support Vector Machines. Machine Learning, 2002, 46(1): 389-422. [11] CHOU C A, KAMPA K, MEHTA S H, et al. Voxel Selection Framework in Multi-voxel Pattern Analysis of fMRI Data for Prediction of Neural Response to Visual Stimuli. IEEE Trans on Medical Imaging, 2014, 33(4): 925-934. [12] RYALI S, SUPEKAR K, ABRAMS D A, et al. Sparse Logistic Regression for Whole-Brain Classification of fMRI Data. NeuroImage, 2010, 51(2): 752-764. [13] LORBERT A, RAMADGE P J. The Pairwise Elastic Net Support Vector Machine for Automatic fMRI Feature Selection // Proc of the IEEE International Conference on Acoustics, Speech and Signal Processing. Vancouver, Canada: IEEE, 2013: 1036-1040. [14] KAMPA K, MEHTA S, CHOU C A, et al. Sparse Optimization in Feature Selection: Application in Neuroimaging. Journal of Global
Optimization, 2014, 59(2): 439-457. [15] HAXBY J V, GOBBINI M I, FUREY M L, et al. Distributed and Overlapping Representations of Faces and Objects in Ventral Temporal Cortex. Science, 2001, 293(5539): 2425-2430. [16] HANKE M, HALCHENKO Y O, SEDERBERG P B, et al. PyMVPA: A Unifying Approach to the Analysis of Neuroscientific Data. Frontiers in Neuroinformatics, 2009. DOI: 10.3389/neuro.11.003.2009. [17] POLDRACK R A, MUMFORD J A, NICHOLS T E. Handbook of Functional MRI Data Analysis. Cambridge, UK: Cambridge University Press, 2011.
2016, Vol. 29 
