Myelin water imaging using a 32-echo Carr-Purcell-Meiboom-Gill (CPMG) spin-echo sequence involves the separation of the signal from different water environments within a voxel based on their transverse relaxation times. We aimed to validate myelin water fraction (MWF) maps derived from this technique using formalin-fixed post mortem spinal cord of people with MS (pwMS) and healthy control cord in order to compare MWF with corresponding histological indices of myelination.An accurate MR measure of myelin in multiple sclerosis (MS) has long been sought and would be valuable in disease monitoring and evaluation of disease modifying therapies. Commonly used MR indices of spinal cord damage such as cord volume may underestimate the true degree of pathological changes significantly [1]. Myelin water imaging was first demonstrated in human brain in vivo by MacKay et al. [2], and analysed using a non-negative least-squares (NNLS) algorithm. Subsequently, MWF obtained from T₂ decay curve studies in fixed post mortem brain was found to scale with optical density of Luxol fast blue, suggesting MWF may be a useful measure of myelination in vivo [3]. However, other parts of the CNS significantly affected by MS, such as the spinal cord, have not been investigated.Three formalin fixed spinal cords from pwMS (1 male, 2 female; age 67-87 years; disease duration 8-44 years; duration of fixation 995-1309 days) and one control (male; age 89; duration of fixation 1320 days) were dissected into 10 cm samples and placed into universal tubes (diameter: 1.5cm). MS spinal cords were imaged on a 3T Philips Achieva using a 3D CPMG sequence (32 echoes, TE n × 10.26ms, TR 1000ms, 0.6×0.6×2mm3 matrix, 20 slices). The control cord was imaged using parameters 32 echoes, TE n × 10ms, TR 3000ms, 0.6×0.6×2mm3 matrix, 21 slices). MWF water maps were calculated using the NNLS algorithm with stimulated echo correction [4], using bounds of 5 – 30ms for short T₂ and 30 – 300ms for intermediate T₂ values. Cords were dissected into blocks of 5mm thickness, processed for embedding in paraffin, and sections stained for Myelin Basic Protein (MBP). Focal areas of complete myelin loss were identified as demyelinated lesions in the white (WML) and grey matter (GML).Figure 1B shows an example curve of the signal decay in one pixel as a function of the echo time. Although the distribution of T₂ showed shortened values relative to previous in vivo studies (short T₂, 10 – 30ms, intermediate T₂, 70 – 1000ms [3]), the two components considered to be associated with myelin and intra/extracellular water are resolved (Fig 1C). Figure 2 shows matched MBP stained histological sections, T₂-weighted MR images from the first echo and MWF maps. Good visual match was achieved between MBP stained sections and myelin water maps. MRI-derived MWF values are shown in Figure 3. Average values in the white and grey matter of the control cord were 41.9±3.8% and 17.9±2.6% respectively. However, myelin values in the lesional areas are close to zero as expected. There was no statistical difference between NLWM and control WM (P-value 0.31).Despite significant shortening of T₂, good separation of the short T₂ component and good quality MWF maps were achieved using fixed post mortem spinal cord of pwMS and controls. The lack of a statistically significant difference between NLWM and control WM as well as between NLGM and control grey matter may be due to the small sample size in this pilot study. Acquisition of a larger dataset and correlation of this promising MR index of myelin is underway, as is correlation with other histological indices (e.g. axonal count).