Brain Degeneration Mechanism And Machine Learning Recognition For Mild Cognitive Impairment

Mild cognitive impairment(MCI)is an early stage of Alzheimer’s disease(AD),and the main symptoms are cognitive disorder and memory loss.Due to the irreversible damage of fibers in AD,diagnosis for MCI is essential to prevent and slow AD occurrence.The clinical symptoms of MCI are mainly caused by the disconnection of fiber pathways whose white matter degenerates.Diffusion tensor imaging(DTI)is an effective method to detect the microstructure changes of white matter.Machine learning provides a systematic and objective auxiliary diagnosis method for MCI recognition.There are mainly three deficiencies in current researches of mechanism and recognition for MCI:(1)white matter degenerates during MCI remains unclear.(2)The relationships between MCI white matter degeneration and behaviors of cognition and emotions are unclear.(3)Unsatisfactory accuracy for MCI recognition urgently needed a feature extraction strategy.In response to these shortcomings,this dissertation has carried out three studies.(1)Quantitative analysis of the differences in white matter fiber tracts between MCI and normal control(NC).(2)Pearson coefficients were calculated between behavior scores and DTI indexes of every fiber.(3)Extract features from the MCI white matter connection network driven by hippocampus related brain regions to improve classification performance.Three innovative results have been obtained:(1)Hippocampus-temporal and thalamus related fibers are found degenerated in MCI.(2)Hippocampus-temporal and thalamus related fibers are correlated with cognitive impairment and depression in MCI.(3)Features extraction of white matter connection network driven by hippocampus related brain regions can improve classification performance significantly.This research provides new evidence for the MCI diagnosis using white matter as a biomarker.Using DTI to explore white matter degeneration in MCI and combining with the neuropsychological test,MRI guidance for early diagnosis of MC could be provided.Appropriate treatment and early intervention for patients could be given to delay the disease progression and reduce the incidence of dementia.