面孔重复抑制的fMRI快速因果模型选择OA

This paper focuses on the causal modulation of brain area connections and the computational efficiency of algorithms for the problem of repetition suppression(RS)in face recognition.It focuses on the modulation effect of immediate and delayed repetitions on the forward and backward connections of brain areas,and proposes a fast dynamic causal model selection(DCMS)algorithm.The computational complexity is greatly reduced by sparse variational inference and linear regression modeling,and the random effect analysis method is used to maintain the model selection performance similar to the traditional method.The experimental data comes from the public data of the open functional magnetic resonance imaging(openfMRI)library.The results show that the computational time of the proposed algorithm is only 6%～10%of the traditional method,but the selected model is consistent with the traditional algorithm;and it is found that when the interference caused by the modulation effect of face perception and recognition is reduced,the modulation effects of immediate repetition and delayed repetition are similar,and there is only a difference in intensity.This is different from the traditional view,indicating that the brain adapts to information processing needs by flexibly adjusting the connection strength at different time scales by the bidirectional feedback and regulation of neurons,rather than relying on independent subsystems.To sum up,the proposed algorithm provides a new idea for the treatment of cognitive disorders.