Nerve fibre pathways and nuclei of the brainstem play an integral role in many functions of the body across different species. In experimental models of spinal cord injury (SCI), trauma-induced plasticity of the brainstem remote from the lesion site could be related to functional recovery¹, while this is less established following SCI in humans. The aim of this study was to assess volumetric and microstructural changes in the brainstem by quantitative MRI and their relation to clinical outcomes in human SCI.30 patients suffering from chronic SCI (>6 months) underwent a comprehensive clinical assessment. 23 healthy subjects were included as control subjects. A multi-parameter mapping quantitative MRI protocol² was applied to all participants. Data were acquired on a 3T Verio or Skyra scanner (Siemens Healthcare). Within the framework of SPM12 (Wellcome Trust Centre for Neuroimaging, UCL), MT-weighted, PD-weighted (PDw), and T1-weighted FLASH volumes were used to calculate quantitative maps of magnetisation transfer saturation (MT) and longitudinal relaxation rate (R1)^2,3. UNICORT was used to estimate and correct bias effects of radiofrequency transmit field inhomogeneity in R1 maps^2,4. Bias field correction for PDw volumes was performed using unified segmentation⁵. The brainstems were then extracted using an in-house atlas for labelling. Eight tissue probability maps were derived through multiparametric brainstem segmentation using a modified multivariate mixture of Gaussians⁶ from a longitudinal dataset (baseline, 6, 12, and 24 months) of MT and R1 maps and bias field corrected PDw volumes of 13 patients and 16 healthy controls (Figure 1). These tissue probability maps were used to segment each individual subject with multichannel unified segmentation⁵ of MT and R1 maps, and PDw volumes. Two Gaussians were chosen for each of the 7 within-brainstem components and 8 Gaussians for the non-brainstem tissue. Geodesic shooting⁷ was used to create a common template and to estimate the Jacobians for tensor-based morphometry (TBM) analysis and to warp the MT and R1 maps into the common template space for voxel-based quantification (VBQ)^2,3 analysis in order to investigate the cross-sectional differences in volume and microstructure. Structural changes in patients compared to controls and between tetra- and paraplegic patients were assessed by two-sample t-tests within SPM12 using explicit masks for each segmented subspace. Initially, a family-wise error corrected voxel threshold of p=0.05 was applied on all statistical parametric maps. Only clusters surpassing a family-wise error corrected cluster threshold of p=0.05 are reported to account for multiple comparisons. Regression models were performed in Stata 13 (StataCorp, College Station, TX) to assess correlations between structural changes and clinical outcomes. All findings are adjusted for age, gender, total intracranial volume, and scanner effects.Fifteen patients suffered from a tetraplegia (six with complete lesion) and 15 from paraplegia (eight with complete lesion), respectively. Mean time since injury was 3.02 years (SD 5.4, Min-Max 0.7-23.8). Mean age was similar between patients (44.7 years, SD 16.7) and controls (36.9 years, SD 11.8). In patients, TBM revealed tract-specific volumetric reductions in the lateral corticospinal tracts (left: p=0.039; right: p=0.044, FWE-corrected) at the midbrain and in the left inferior cerebellar peduncle (p=0.044, FWE-corrected) at the border between pons and medulla oblongata. Volumetric reductions in brainstem nuclei was found in the right substantia nigra (p=0.039, FWE-corrected) at the midbrain level. No differences were found when comparing tetraplegic versus paraplegic patients. Correlations between Spinal Cord Independence Measure (a functional outcome measure of daily living with lower scores indicating more severe impairment) and volumetric changes in the inferior cerebellar peduncle (R2=0.169, p=0.027, Figure 3A) and left lateral corticospinal tract (R2=0.230, p=0.008, Figure 3B) were found.This study demonstrates focal changes affecting specific pathways and nuclei involved in transmitting and processing of motor information within the brainstem following SCI using multiparametric segmentation and a quantitative multi-parameter mapping protocol. Atrophic changes and their relation to clinical outcomes were already shown within the brain and spinal cord in acute and chronic patients^8,9, but such findings are less established in the brainstem¹⁰. In a next step, VBQ of myelin-sensitive MT and R1 will be applied next to better understand pathology and the underlying changes in microstructure.SCI affects the whole central nervous system from the spinal cord to the brainstem and brain. Our findings demonstrate that SCI affects specific pathways and nuclei in the brainstem and that these changes are associated with functional outcome. Neuroimaging surrogate markers seem feasible to reveal potential mechanisms of interventions and could be applied to complement clinical assessments in rehabilitation and clinical trials.