Design On Nuclear Magnetic Resonance Database System And Research On Implementation Of An Analytical Diagnostic For Alzheimer's Disease

Alzheimer's disease has no obvious specific biological markers.In clinical practice,Alzheimer's disease examination methods mainly include some neuropsychological tests and imaging examinations by means of magnetic resonance imaging.Therefore,in order to accurately diagnose the patient,the hospital generally collects many items of examination data.In order to manage these data,this paper uses computer technology to establish a patient information management system to manage these complex data.With the maturity and wide application of machine learning algorithms,this paper establishes a classification model of auxiliary diagnosis through relevant algorithms to help doctors diagnose the disease in some complicated data situations and provide reference.Firstly,this paper combines the research of medical information systems implemented by other researchers,analyzes the needs of different people who need data management systems,designs and analyzes the systems according to the requirements analysis,improves the system structure,and builds the environment for the development of the system.Then,the programming of Alzheimer's disease auxiliary diagnosis system was realized.Each table field of the database system was designed specifically,each function module of the system was perfected,and user information management,patient basic information query management,and magnetic resonance image inquiry were realized.Multi-criteria screening and patient data upload capabilities.At the system development stage,the system's security was considered and specific solutions were made.Finally,using the original data of the ADNI database as a training set,Alzheimer's disease classification algorithm was designed and an auxiliary diagnosis model was implemented.Three Alzheimer's disease classifier models were designed as controls.First,the downloaded data was sorted out,and the original data was cleaned.This made up for the lack of original data and duplication.By Naive Bayesian,multi-layer perceptron neural network,and support vector machine using Gaussian kernel function,these three methods establish a classifier for comparison,and calculate the corresponding accuracy rate by 10-fold cross validation method.The original training set was reduced by the principal component analysis,and then the PCA-Adaboosting Alzheimer's assistant diagnosis classifier was established through the Adaboosting algorithm based on the CART decision tree.