|
|
|
| Joint and Deep Ensemble Regression of Clinical Scores for Alzheimer′s Disease Using Longitudinal andIncomplete Data |
| Yang Mengya1, Hou Wen2, Yang Peng1, Zou Wenbin2, Wang Tianfu1, Lei Baiying1* |
1.(School of Biomedical Engineering, Health Science Center, Shenzhen University, National-Regional Key Technology Engineering Laboratory for Medical Ultrasound, Guangdong Key Laboratory for Biomedical Measurements and Ultrasound Imaging, Shenzhen 518060, Guangdong, China)
2.(School of Information Engineering, Shenzhen University, Shenzhen 518060, Guangdong, China) |
|
|
|
|
Abstract Alzheimer′s disease (AD) is a neurodegenerative disease with an irreversible and progressive process, and thus close monitoring is essential for making adjustments in the treatment plan. Since clinical scores can indicate the disease status effectively, the prediction of the scores based on the magnetic resonance imaging (MRI) data is highly desirable. Different from previous studies at a single time point, we proposed to build a model to explore the relationship between MRI data and scores, thereby predicting longitudinal scores at future time points from the corresponding MRI data. The model incorporated three parts, correntropy regularized joint learning based features election, deep polynomial network based feature encoding and finally, support vector regression. The regression process was carried out for two scenarios. One was desirable in practice, which is to use baseline data for predictions at future time points, and the other was to further improve the prediction accuracy, which was to combine all the previous data for the prediction at the next time point. Meanwhile, the missing scores were filled in the second scenario to address the incompleteness presented in the data. We predicted longitudinal scores at future time points by the proposed model. Besides, the corresponding average absolute error and Pearson correlation were calculated to estimate the experimental results. In scenario 1, the average absolute value was 2.01, 2.06, 2.06, 2.27, 2.00 and the Pearson correlation coefficient was 0.70, 0.69, 0.56, 0.65, 0.67. In the scenario 2, the average absolute error was 0.14, 0.10, 0.09, 0.08 and the Pearson correlation coefficient is 0.72, 0.75, 0.78, 0.74. The simulation results validated that the proposed model described accurately the relationship between MRI data and scores, and thus was effective in predicting longitudinal scores.
|
|
Received: 16 October 2017
|
|
|
|
|
|
[1] Petersen R C, Doody R, Kurz A, et al. Current concepts in mild cognitive impairment [J]. Archives of Neurology, 2001, 58(12): 1985-1992.
[2] Castellani R J, Rolston Raj K, Smith M A. Alzheimer disease [J]. Disease-A-Month: DM, 2010. 56(9): 484-494.
[3] Bain L J, Jedrziewski K, Morrison-Bogorad M, et al. Healthy brain aging: A meeting report from the sylvan m. cohen annual retreat of the university of pennsylvania institute on aging [J]. Alzheimer′s & Dementia, 2008. 4(6): 443-446.
[4] Brookmeyer R, Johnson E, Ziegler-Graham K, et al. Forecasting the global burden of Alzheimer′s disease [J]. Alzheimer′s & Dementia, 2007, 3(3): 186-191.
[5] Prince M, Comas-Herrera A, Knapp M, et al. World Alzheimer report 2016: Improving healthcare for people living with dementia: coverage, quality and costs now and in the future [EB/OL]. https://www.alz.co.uk/research/world-report-2016, 2016-09/2017-10-16.
[6] Pich E M, Jeromin A, Frisoni GB, et al. Imaging as a biomarker in drug discovery for Alzheimer′s disease: Is MRI a suitable technology? [J]. Alzheimer′s Research & Therapy, 2014. 6(4): 51.
[7] Zhu Xiaofeng, Suk H, Lee SW, et al. Subspace regularized sparse multitask learning for multiclass neurodegenerative disease identification [J]. IEEE Transactions on Biomedical Engineering, 2016. 63(3): 607-618.
[8] Jie Biao, Liu Mingxia, Liu Jun, et al. Temporally constrained group sparse learning for longitudinal data analysis in Alzheimer′s disease [J]. IEEE Transactions on Biomedical Engineering, 2017. 64(1): 238-249.
[9] Zhu Xiaofeng, Huang Zi, Shen Hengtao, et al. Dimensionality reduction by mixed kernel canonical correlation analysis [J]. Pattern Recognition, 2012, 45(8): 3003-3016.
[10] Wang Kaiye, He Ran, Wang Liang, et al. Joint feature selection and subspace learning for cross-modal retrieval [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016. 38(10): 2010-2023.
[11] Wang Xiaoqian, Shen Dinggang, Huang Heng. Prediction of memory impairment with MRI data: A longitudinal study of Alzheimer′s disease [C]// International Conference on Medical Image Computing and Computer-Assisted Intervention. Shanghai: Springer, 2016: 273-281.
[12] Zhang Daoqiang, Liu Jun, Shen Dinggang. Temporally-constrained group sparse learning for longitudinal data analysis [J]. Medical Image Computing and Computer-Assisted Intervention-MICCAI 2012, 2012: 264-271.
[13] Zhang Daoqiang, Shen Dinggang, Initiative Alzheimer′s Disease Neuroimaging. Multi-modal multi-task learning for joint prediction of multiple regression and classification variables in Alzheimer′s disease [J]. NeuroImage, 2012. 59(2): 895-907.
[14] Livni R, Shalev-Shwartz S, Shamir O. An algorithm for training polynomial networks [J]. arXiv preprint:1304.7045, 2013.
[15] Shi Jun, Zheng Xiao, Li Yan, et al. Multimodal neuroimaging feature learning with multimodal stacked deep polynomial networks for diagnosis of Alzheimer′s disease [J]. IEEE Journal of Biomedical and Health Informatics, 2017,22:173-183.
[16] Liu Xiao, Shi Jun, Zhang Qi. Tumor classification by deep polynomial network and multiple kernel learning on small ultrasound image dataset [C] //International Workshop on Machine Learning in Medical Imaging.Berlin: Springer, 2015: 313-320.
[17] Shi Jun, Zhou Shichong, Liu Xiao, et al. Stacked deep polynomial network based representation learning for tumor classification with small ultrasound image dataset [J]. Neurocomputing, 2016, 194: 87-94.
[18] Greenspan H, Van Ginneken B, SummersR M. Guest editorial deep learning in medical imaging: Overview and future promise of an exciting new technique [J]. IEEE Transactions on Medical Imaging, 2016, 35(5): 1153-1159.
[19] Huang Lei, Jin Yan, Gao Yaozong, et al. Longitudinal clinical score prediction in Alzheimer′s disease with soft-split sparse regression based random forest [J]. Neurobiology of Aging, 2016, 46: 180-191.
[20] Suk H, Lee S, Shen Dinggang, et al. Deep ensemble learning of sparse regression models for brain disease diagnosis [J]. Medical Image Analysis, 2017, 37: 101-113.
[21] Lei Baiying, Yang Peng, Wang Tianfu, et al. Relational-regularized discriminative sparse learning for Alzheimer′s disease diagnosis [J]. IEEE Transactions on Cybernetics, 2017, 47(4): 1102-1113.
[22] Zhu Xiaofeng, Li Xuelong, Zhang Shichao, et al. Robust joint graph sparse coding for unsupervised spectral feature selection [J]. IEEE Transactions on Neural Networks and Learning Systems, 2017, 28(6): 1263-1275.
[23] Tibshirani R, Saunders M, Rosset S, et al. Sparsity and smoothness via the fused lasso [J]. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 2005, 67(1): 91-108.
[24] Nesterov Y. Gradient methods for minimizing composite objective function[J]. Mathematical Programming, 2013,140(1): 125-161.
[25] Gaonkar B, Shinohara RT, Davatzikos C, et al. Interpreting support vector machine models for multivariate group wise analysis in neuroimaging [J]. Medical image analysis, 2015, 24(1): 190-204.
[26] Sluimer Jasper D, van der Flier Wiesje M, Karas Giorgos B, et al. Accelerating regional atrophy rates in the progression from normal aging to Alzheimer′s disease [J]. European Radiology, 2009, 19(12): 2826.
[27] Misra C, Fan Yong, Davatzikos C. Baseline and longitudinal patterns of brain atrophy in MCI patients, and their use in prediction of short-term conversion to AD: Results from ADNI [J]. Neuroimage, 2009, 44(4): 1415-1422.
[28] Yuan, Ming, Yi Lin. Model selection and estimation in regression with grouped variables[J]. Journal of the Royal Statistical Society: Series B, 2006, 68(1): 49-67. |
|
|
|
