Design And Implementation Of Dbs Preoperative Evalution System For Pd Based On Image Diagnosis

Parkinson’s Disease(PD)has become the second most common neurodegenerative disease in the world,bringing suffering to patients and causing global economic losses.Deep Brain Stimulation(DBS)is a common therapy for the treatment of PD at home and abroad.Two versions of DBS treatment consensus have been produced in China for competent surgical teams,in which preoperative evaluation is an important part of the therapy.According to the treatment consensus,preoperative evaluation of DBS must include clear diagnosis of PD,assessment of motor symptoms and non-motor symptoms.The current preoperative evaluation of DBS using paper or traditional office software is cumbersome,and patient data is difficult to manage and easily lost in this way.There are many evaluation regulations based on the gold standard scale for PD diagnosis,and only digitizing it cannot actually improve the situation where diagnosis work is difficult to carry out.The use of neuroimaging combined with deep learning techniques is now a promising avenue for the task of classifying PD patients from healthy populations.Therefore,this thesis builds a preoperative evaluation system for DBS using artificial intelligence technology to assist in the diagnosis of Parkinson’s disease,based on software engineering specifications,in order to improve the convenience and reliability of physicians’ evaluation work.The main work of this thesis focuses on the diagnosis of PD in preoperative evaluation.Firstly,through literature research,the noninvasive,radiation-free and affordable Magnetic Resonance Imaging was selected as the subject of the study to enhance the practical application of the model.Secondly,the preprocessing work of the public dataset Parkinson’s Progression Markers Initiative(PPMI)was completed by batch processing.Then the construction of Parkinson’s disease classification model based on DenseNet was completed and the model was evaluated using five-fold cross-validation.Finally,according to the practical work of doctors and the two versions of therapy consensus,the requirements of DBS preoperative evaluation system were analyzed,and the system design,implementation and testing were completed.The main contribution of this thesis is to improve the two-dimensional DenseNet pre-training model into a three-dimensional structure,while enhancing the model performance through a cross-dimensional parameter migration method,which is serviced to achieve an intelligent diagnosis of Parkinson’s disease during preoperative assessment.In addition,this thesis introduces microservice components to realize the remote call of Web service to model diagnostic service.The two finally present a dynamically extensible integrated preoperative evaluation system,which can efficiently and reliably assist doctors to complete the patient evaluation work and has certain application value.