Research On Auxiliary Diagnosis Of Alzheimer’s Disease Based On Multiscale Entropy Algorithm

Mild Cognitive Impairment(MCI)is a transitional stage from normal cognitive decline to Alzheimer’s Disease(AD),so monitoring MCI is the key to AD prevention and intervention.Studies have shown that the complexity of Blood Oxygen Level Dependent(BOLD)signaling is an important indicator for studying MCI and neurodegenerative diseases associated with cognitive decline.This paper proposes an improved Multi-Scale Entropy(MSE)algorithm based on the overlapping mean coarse-graining method to study the complexity of BOLD signals and monitor MCI.First,the time series features of 116 brain regions under the Automated Anatomical Labeling(AAL)template were extracted based on DPABI,and entropy analysis was performed on 90 key brain regions at 6 scales.Healthy(HC)subjects,early-stage mild cognitive impairment(EMCI)patients,late-stage mild cognitive impairment(LMCI)patients,and Alzheimer’s disease were obtained from the Alzheimer’s Disease Neuroimaging Initiative(ADNI)database(AD)patients with functional Magnetic Resonance Imaging(f MRI)data.By setting key parameters,DPABI was used to obtain the time series signals of different brain regions of the participants;then the entropy values of different brain regions of the participants at different scales were calculated based on the improved multi-scale entropy algorithm.Second,brain regions with significant differences were extracted based on one-way ANOVA and independent samples t-test.One-way analysis of variance was performed on the entropy results of the four groups of HC,EMCI,LMCI and AD to obtain 11 brain regions with significant differences in the four groups of data,and then the HC and EMCI,HC and LMCI,HC and AD,The entropy results of EMCI and LMCI,EMCI and AD,LMCI and AD six groups of experiments were carried out by independent sample t test,and the brain areas with significant differences under different scale factors were obtained.Finally,a saliency feature extraction method based on the principle of information maximization is proposed,and an AD diagnosis model is established.Taking the entropy values of brain regions with significant differences as features,the method of combining Support Vector Machines(SVM)and N-fold cross-checking was used to conduct MCI diagnosis research.HC,EMCI,LMCI,and AD were classified into four categories and two categories,respectively,and compared with the classification of all brain area data as features,the final result was that the entropy values of brain regions with significant differences were used as features for classification higher accuracy.In this paper,an MSE-based clinical imaging monitoring method for AD is proposed,which reveals the brain regions that change significantly in the process of AD formation,and establishes a monitoring model to provide assistance for clinical diagnosis.