Hybrid Diffusion Imaging (HYDI) in a Brain Model of Dysmyelination

Introduction Diffusion tensor imaging (DTI) is widely used for the study of white matter (WM) diseases and fractional anisotropy (FA) is commonly used as a measure of WM integrity. However, FA is also highly sensitive to such factors as non-Gaussian diffusion, crossing fibers and imaging noise, which may degrade its specificity. It has recently been proposed that the axial diffusivity (D//) of the diffusion tensor is specifically related to axonal integrity while the radial diffusivity (D⊥) is related to myelination [1]. Alternatively, q-space imaging and diffusion spectrum imaging (DSI) may provide additional information about WM microstructure. The zero displacement probability (Po) is a measure of water diffusion restriction [2,3] that has been found to have significant diagnostic value in multiple sclerosis (MS) [2]. In this study, the dysmyelinating shaking (sh) pup model was studied using both the DTI and DSI measurements acquired from a hybrid diffusion imaging (HYDI) approach [3]. The sh pup is a canine mutant with a profound paucity of myelin, without the confounding effects of axonal loss, inflammation or edema [4]. This reductionist disease model may help to disentangle the many confounding pathological changes that occur in MS and other WM diseases and relate them to changes in diffusion properties observable with MR. Materials and Methods Six sh pups and four age-matched control dogs were scanned (once or twice) at ages ranging between 4 and 46 months. HYDI experiments were performed on anesthetized pups with a 3T GE SIGNA scanner with SS-SE-EPI and a quadrature extremity coil. The HYDI sampling scheme (Table 1) consisted of four icosahedral shells. The second shell was used for DTI processing, the whole dataset was used for DSI processing [3]. Other MR parameters were: TR/TE=7500/140ms, matrix= 96x96, FOV=15 cm and 26 axial 3mm slices. DTI and DSI measures including FA, MD, D//, D⊥, Po and mean squared displacement (MSD) were post-processed. WM tissues were segmented by two-stage segmentation using the FAST algorithm [5]. The first 2-class segmentation was done on MD to eliminate CSF from parenchyma; the second 2-class segmentation was done on FA to separate WM and GM. Diffusion measures of WM were compared between controls and sh pups and observed across the age range. In addition to whole brain WM, ROI analysis of the bilateral internal capsules was performed to specifically assess the most compact white matter.