Many studies in the literature have validated the use of resting-state fMRI (rs-fMRI) for brain disorder/disease identification. Unlike the existing methods that mostly first estimate functional connectivity and then extract features with a graph theory, in this paper, we propose a novel method that directly models the temporal stochastic patterns inherent in BOLD signals for each Region Of Interest (ROI) individually. Specifically, we model temporal BOLD signal fluctuation of an individual ROI by means of Hidden Markov Models (HMMs), and then compute a regional BOLD signal likelihood with the trained HMMs. By regarding the BOLD signal likelihood of ROIs over a whole brain as features, we build a classifier that can discriminate subjects with Autism Spectrum Disorder (ASD) from Normal healthy Controls (NC). In addition, we also devise a method to further investigate the characteristics of temporal dynamics in rs-fMRI estimated by HMMs. For group comparison, we use the metrics of state occupancy rate and lifetime of the optimal hidden states that best represent the temporal BOLD signals. In our experiments with ABIDE cohort, we validated the effectiveness of the proposed method by achieving the highest diagnostic accuracies among competing methods. We could also identify the group differences in temporal dynamics between ASD and NC in terms of state occupancy rate and lifetime of individual states.
|Title of host publication||Connectomics in NeuroImaging - 1st International Workshop, CNI 2017 Held in Conjunction with MICCAI 2017, Proceedings|
|Editors||Leonardo Bonilha, Guorong Wu, Paul Laurienti, Brent C. Munsell|
|Number of pages||9|
|Publication status||Published - 2017|
|Event||1st International Workshop on Connectomics in NeuroImaging, CNI 2017 held in conjunction with the 20th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2017 - Quebec City, Canada|
Duration: 2017 Sept 14 → 2017 Sept 14
|Name||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|
|Other||1st International Workshop on Connectomics in NeuroImaging, CNI 2017 held in conjunction with the 20th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2017|
|Period||17/9/14 → 17/9/14|
Bibliographical noteFunding Information:
Acknowledgement. This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2015R1C1A1A01052216) and also partially supported by Institute for Information & Communications Technology Promotion (IITP) grant funded by the Korea government (No. 2017-0-00451).
© 2017, Springer International Publishing AG.
ASJC Scopus subject areas
- Theoretical Computer Science
- Computer Science(all)