Study On High-resolution Quantitative Susceptibility Mapping For Deep Brain Stimulation In Parkinson’s Disease

Parkinson’s disease(PD)is a movement disorder disease.Its typical clinical manifestations are tremor,rigidity and bradykinesia.Deep brain stimulation(DBS)is an important treatment for PD.However,due to the heterogeneity and subjectivity of individualized diagnosis and treatment of PD,the precise positioning of surgical targets and the evaluation of surgical benefit rates are still challenging.Quantitative susceptibility mapping(QSM)is an image reconstruction technology that can quantitatively detect the distribution of iron deposits associated with PD pathology.The QSM images with isotropic submillimeter voxel size can well display the fine deep gray matter nuclei.Therefore,high-resolution QSM is a promising imaging technology for accurate localization of target nuclei and evaluation of surgical benefit rate.This thesis focused on the application of high-resolution QSM in the field of PD-DBS in the following three aspects:(1)Research on the post-processing method of susceptibility-magnitude fusion image forDBS targetingThe purpose of this study was to develop an image post-processing method to obtain susceptibility-magnitude fusion images for the accurate PD-DBS targeting.A total of 25 PD patients were included in this study and a 3T magnetic resonance imaging(MRI)scanner was used to obtain high-resolution QSM images with a voxel size of 0.90×0.90×0.90 mm~3.To create the susceptibility-magnitude fusion images,the high-resolution QSM images of the STN region were extracted by a spherical mask and fusion images were created by fusing the magnitude images of the whole brain and the extracted QSM images.The fusion images were finally obtained by linear adjustment of the intensity range of the magnitude and QSM images.Compared with T2-weighted images,the fusion images yielded a higher visual score,inter-rater consistency of visual score,and contrast noise ratio for STN depiction.Furthermore,the skull in the images can be used in the DBS stereo positioning navigation system.The susceptibility-magnitude fusion images generated by the proposed image post-processing method are helpful for the precious PD-DBS targeting.(2)Exploring the relationship between the DBS benefit and the spatial distribution ofiron in deep gray matter nucleiThe purpose of this study was to explore the relationship between the spatial distribution of iron deposition in deep brain gray matter nuclei and DBS surgical outcomes in PD patients.A total of 40 PD patients were included in this study.Motor scores were first assessed before surgery and a 3T MRI scanner was used to obtain high-resolution QSM images.Texture analysis was used to calculate texture features that can reflect the spatial distribution of iron in the substantia nigra(SN),STN and dentate nucleus(DN).Postoperative motor scores were obtained at 6-months follow-up.Regression analyses were used to explore the relationship between the texture features and the motor improvement ratio.Results showed that the motor improvement after STN-DBS was found to be correlated with the second-order texture parameters in the SN and correlated negatively with the mean susceptibility value in the DN.The results of this study indicate that the spatial distribution of iron in the gray matter nucleus of PD patients plays an important role in the prognosis of DBS surgery.This study is the first time to explore iron deposition related to PD pathology and reveals that the spatial distribution of iron in the gray matter nuclei of the deep brain is a potential marker for predicting the DBS response.(3)Automatic segmentation of midbrain nuclei based on high-resolution QSM image and convolutional neural networkThis study aims to segment midbrain structures in high-resolution susceptibility maps using a method based on a convolutional neural network(CNN).The susceptibility maps of 75subjects were acquired with a voxel size of 0.83×0.83×0.80 mm~3on a 3T MRI scanner to distinguish the red nucleus(RN),SN,and STN.A deeply supervised attention U-net was pre-trained with a dataset of 100 subjects containing susceptibility maps with a voxel size of 0.63×0.63×2.00 mm~3 to provide initial weights for the target network.Results showed that the proposed segmentation strategy could extract the RN,SN and STN accurately and yielded comparable results to manual delineation.The overall volume and magnetic susceptibility values of the nuclei extracted by the automatic CNN method were significantly correlated with those by manual delineation.This transfer learned CNN allows excellent segmentation of deep gray nuclei on QSM images and provides a supplementary means for automatic localization of the target nucleus in DBS surgical and quantitative analysis of QSM research.In conclusion,high-resolution QSM can clearly display the boundary of the midbrain gray matter nucleus and obtain the distribution of iron deposits in the deep gray matter nucleus.It is helpful for the accurate localization of the target nucleus and the evaluation of the surgical benefit during PD-DBS surgical treatment.This thesis provides a new research perspective to aid clinical decision-making and prognosis assessment.Meanwhile,it also shows the potential of high-resolution QSM as a key tool in the precision treatment of DBS in the future.