Detecting Change Points in fMRI Data via Bayesian Inference and Genetic Algorithm Model.

Xiuchun Xiao,Bing Liu,Jing Zhang,Xueli Xiao,Yi Pan
DOI: https://doi.org/10.1007/978-3-319-59575-7_28
2017-01-01
Abstract:Dynamic functional connectivity detection in fMRI has been recently proved to be powerful for exploring brain conditions, and a variety of methods have been proposed. This paper mainly investigates the field of change point detection based on Bayesian inference and genetic algorithm. We define different indicator vectors as different individuals, which represent some possible change point distributions, and use Bayesian posterior probability to evaluate their fitness. Accordingly, we also present an improved genetic algorithm, which is applied to evolve the individuals toward the best one. Then, the most possible change points distribution could be resolved. The method has been applied to several synthesized data and simulation results reveal that the proposed method can detect change points in fMRI datasets with higher precision and lower time consumption.
What problem does this paper attempt to address?