Abnormal Brain Function Network Analysis Of Alzheimer’s Disease Based On Improved ICA

With the continuous development of Resting-state Functional Magnetic Resonance Image(RS-fMRI)technology,the image processing technology is also gradually mature.For patients with Alzheimer’s disease(AD),effectively separating meaningful neural signals from RS-fMRI images and constructing a brain function network have important research significance for their earlier disease prediction.Independent Component Correlation Algorithm(ICA)is one of the widely used signal extraction algorithms in the field of RS-fMRI image processing,but there are still shortcomings.Therefore,this article makes an effective improvement for the traditional ICA,and will apply it to the analysis of brain networks of Alzheimer’s patients.Through experimental analysis and comparison,it effectively illustrates the effectiveness and robustness of the improved algorithm proposed in this paper.The main contributions of the paper are as follows:Based on the classical ICA method,the noise terms of the data will be assumed,so that it will be expanded and analyzed under the original assumed non-noise condition.The robustness of ICA will be improved by adjusting the main components of the noise,thus implementing the Confounding-robust Independent Component Algorithm,CRICA.The proposed method allows the existence of static confounding noise,which can adjust the different noise in each group to merge the changes between the groups together.It is robust to prevent mixing within the group and can be used for data analysis across groups.The improved algorithm CRICA is compared with FastICA and GroupICA algorithms by using simulated data and real data,respectively.The results show that the CRICA method performs well in models with group confounding and no confounding.The CRICA method can enhance the robustness of ICA by adjusting group confusion.Therefore,when the data contains fixed noise or confusion,CRICA will get better results and reach the expected judgment.This paper applies the CRICA algorithm to Alzheimer’s brain network research.For RS-fMRI image data,this paper uses CRICA to study the statistical differences in the resting state networks of 30 healthy controls(HC),19 patients with amnestic mild cognitive impairment(aMCI),and 22 patients with AD.After using the CRICA method,the functionally connected abnormal areas analyzed in this paper are consistent with previous studies.Changes in functional connectivity were found in areas medial superior frontal gyrus(SFGmed),dorsolateral superior frontal gyrus(SFGdor.L),and Bilateral calcarine fissure and surrounding cortex(CAL)that were different but consistent with clinical presentation.