Diffusion MRI-based Study In Focal Cortical Dysplasia And Visual White Matter

Diffusion magnetic resonance imaging(dMRI)can measure the diffusivity of water molecules within the brain tissues and is an effective method to evaluate the microstructural properties of tissues.Moreover,dMRI is the only widely used noninvasive imaging technology that allows the visualization of white matter(WM)fiber bundles.Even so,the potential of dMRI in clinical applications and pre-clinical researches has not been fully explored,and there are still many application scenarios worth exploring.Therefore,the present research work attempts to use dMRI-based microstructural features and WM fiber tracking technology to provide some novel research perspectives for clinical and pre-clinical issues.This research work focuses on:(1)exploring whether dMRI-based microstructural features at voxel levels can improve the detection of focal cortical dysplasia(FCD)lesions;(2)investigating the microstructural properties of major WM bundles in patients with strabismus before and after surgery using dMRI fiber tracking techniques;(3)Combining dMRI fiber tracking and blood oxygen level-dependent(BOLD)functional magnetic resonance imaging(fMRI)to explore WM functional significance of optic radiation(OR)sub-bundles.The first study:automated detection of FCD lesions based on dMRI.An accurate detection of FCD lesions is the key to successful resection.This study used voxel-based dMRI microstructural features and T2-weighted image(T2WI)intensity features to implement automated classification and recognition of FCD lesions to explore the contribution of dMRI in the automated detection of FCD lesions.The 1.5T MRI data of 12 FCD patients was applied to train and cross-validate GPML(i.e.,Gaussian processes for machine learning)and SVM(i.e.,support vector machine)classifiers.In addition to that,the performances of multi-modal GPML versus single-modal GPML were compared.The results demonstrated that,compared with SVM(AUC=0.69),the proposed GPML(AUC=0.76)paradigm showed better performance and was robust to unbalanced datasets.Meanwhile,the GPML performance on dMRI-based features(AUC=0.63)was virtually identical to that of classifiers on T2WI intensity features(AUC=0.64).Moreover,the performance of GPML was significantly improved when applied to multi-modal features,inclusive of dMRI-based features.The above results suggest that the joint combination of multi-modal MRI data,inclusive of dMRI,and GPML,significantly improve the accuracy of FCD lesion detection,providing a promising alternative for detection of FCD lesions.The second study:microstructural properties of major WM tracts in strabismus before and after strabismus surgery.This study,using dMRI,explored the neurological mechanism of postoperative visual function improvement in patients with strabismus.This study acquired dMRI data from patients with constant exotropia,before(n=19)and after surgery(n=15),and healthy controls(n=20),and evaluated ocular dominance(OD)and stereopsis.24 WM bundles were constructed and their mean diffusivity(MD)and fractional anisotropy were computed.Microstructural parameters estimated from these bundles were further assessed for group differences.Then,we evaluated the correlation between OD and the absolute value of microstructural changes.The results showed that,after the operation,all patients restored to normal stereopsis,but the OD of the most of the patients was still abnormal(9 out of 11).Compared with that of preoperation,the MD of the postoperative patient reduced significantly along the left anterior thalamic radiation(ATR),left arcuate fasciculus(AF),left corticospinal tract(CST),left cingulum cingulate(CGC)and left inferior fronto-occipital fasciculus.Moreover,OD was negatively correlated with the absolute values of MD alterations in left ATR,left AF,left CST,and left CGC.These results indicated that the microstructural changes after surgery in the visualspatial fiber bundles may be related to the restoration of stereopsis.On the other hand,the results of correlation analysis may illustrate that balanced binocular input is more conducive to the recovery and improvement of binocular visual functions including stereopsis.The third study:population receptive field(pRF)estimation in sub-bundles of human OR.While functional activity of brain WM has been reliably detected,the functional significance of OR sub-bundles,reconstructed and segmented based on dMRI,has not been explored yet.In this study,7T retinotopy datasets from Human Connectome Project were used to reconstruct the OR.Specifically,OR segmention into sub-bundles was performed according to the retinotopic map of the primary visual cortex(V1).To suppress the influence of gray matter(GM)signals,WM masks set to different degrees of restraint were used.The OR sub-bundles were confined within the spaces of these masks,which were ranked as level 1 sub-bundles,level 2 sub-bundles,and level 3 sub-bundles.The pRF model was then applied to evaluate the retinotopic properties of these sub-bundles.Moreover,the consistency of the pRF properties of level 1 sub-bundles with V1 sub-fields at the endpoint of the sub-bundles were evaluated.Correlation analysis was performed to test the relationships of the pRF parameters of level 2 and 3 sub-bundles with those of the level 1 sub-bundles.The results showed that the pRF properties of differentialy constrained sub-bundles exhibited the standard retinotopic properties,and the pRF properties of the level 1 subbundles were in good agreement with those of V1 sub-fields.Moreover,the pRF parameters of level 2 and 3 sub-bundles were significantly correlated with level 1-pRF parameters,indicating that OR sub-bundle evaluation was GM-free.These findings suggest that the BOLD signal of the OR sub-bundles plays a key role in encoding highfidelity visual information.In summary,this research shows that dMRI combined with a specific paradigm provides novel research perspectives for the automated lesion detection of FCD,postoperative neural mechanism of stereopsis restoration,and exploration of functional significance of white matter,which eventually promotes the research progress in related fields.