Research On Computer Aided Diagnosis Model For Early Alzheimer's Disease By Means Of Multi-modal

Alzheimer's disease(AD)is one of the most difficult diseases to cure at present,which seriously affects the normal life of the elderly and their families.Mild Cognitive Impairment(MCI)is a state of early AD and may be converted to AD later.MCI is often mistaken for the normal aging and misses the best time of treatment.Therefore,accurate diagnosis of MCI is essential for the early detection and early treatment of AD.This paper proposes a deep learning model for early diagnosis of AD that simulates the clinician's diagnostic process.Neuroimaging is an important method for diagnosing structural and functional changes in the brain of patients,and plays an important role in the early diagnosis and prediction of AD.In this paper,the PET and MRI images commonly used in clinical diagnosis of AD are used to train a model with high accuracy of PET and MRI images through convolutional neural networks.First,pre-training is performed on the similar dataset to migrate the trained model parameters to the new network.On the ADNI(Alzheimer's Disease Neuroimaging Initiative,ADNI)dataset,two independent convolutional neural networks(VGGNet-16)were used,and the original activation function ReLU of the network was improved by using the PReLU function to learn PET images and MRI image features respectively,trained AD imaging diagnostic model.In this paper,without changing the sample category,the image is processed by rotation,scaling,grayscale contrast adjustment and so on.to expand the dataset to make up for the problem of fewer samples of different poses.Data amplification is used to increase the amount of data and solve the problem of imbalance between classes.And the comparative analysis of the data set before and after data amplification shows that the method of data amplification is used to play a certain regularization effect and improve the recognition effect of the model.Different modal data reflect different pathologies.The diagnosis of AD needs to be combined with different diagnostic methods to obtain accurate diagnosis results.Single modal data is not representative in information representation,and multimodal data is often used for comprehensive diagnosis.When clinicians diagnose AD,they usually refer to a variety of neuroimaging findings.In this paper,MRI images,PET images,MMSE scales and CDR scale clinical cognitive test data are combined.Firstly,MRI and PET imaging diagnostic models with high recognition accuracy are trained,and then multi-modal medical images are analyzed by correlation analysis.The auxiliary diagnosis results were combined with the clinical evaluation scale,and finally the comprehensive evaluation results of the patient's pathology and psychology were integrated.Experiments on the data set show that the proposed method improves the accuracy of diagnosis and is superior to other aided diagnostic models in many indicators,providing new ideas for multimodal diagnosis of Alzheimer's disease.