Equine brain template and atlas :
Equisobrain dataset :
We compared neurological, physiological and behavorial trajectories of young horses benefiting either of a prolonged maternal presence (N=12) or separated from their mother half-way through childhood, at 6 months of age (N=12). All animals lived in the same rich social groups and have been raised in accordance with the European directive 2010/63/EU for animal protection and welfare used for scientific purposes. We applied a holistic approach included an MRI-based whole-brain exploration for the first time in this species. We showed the multi-faceted positive effect of maternal presence beyond the early post-natal phase. Notably, we provide pioneering evidence of the maternal presence’s impact on neuroanatomical, microstructural and basal activity development of critical brain regions (posterior cortical areas and the hypothalamus) and the Default Mode Network (DMN) connectivity development. Furthermore, our study characterised a distinctive and consistent developmental trajectory for the juveniles benefiting from the presence of their mother, associated with a more positive and adaptive social behaviour, and improved feeding behaviour and physical development on a long-term basis. Beyond demonstrating the holistic benefits of a prolonged maternal presence for young, long-developing mammals, our study also lays the groundwork for an innovative model in the investigation of neurobiological development using non-invasive neuroimaging compatible with longitudinal studies.
Subjects were 24 Welsh foals, 12 females and 12 males, 6 months old and their 24 mothers 7 years old. All horses (foals and mothers) were born and kept at the Animal Physiology Experimental Unit PAO (UEPAO, 37380 Nouzilly, France), INRAE. They all lived in groups in, depending on the season, indoor collective stalls (20 m x 25 m) on straw with free access to an outdoor paddock, or in large outdoor grass pastures. Hay and/or grass and water were available ad libitum. The foals were born in May and June 2020. During the first five months of their life, they were all kept with their mothers in the same outdoor pasture with two more mother-fillies dyads born and kept under the same conditions. In October 2020, two distinct groups were formed and kept in two different pastures: 12 dyads of male foals/mothers (Group A), and 12 dyads of female foals/mothers (Group B). All procedures were performed in accordance with the European directive 2010/63/EU for animal protection and welfare used for scientific purposes and approved by the local ethical committee for animal experimentation (CEEA VdL, Tours, France) and animals have been rehabilitate after the experimentation.
Imaging procedures have been realized in PIXANIM imaging facilities, on the same site as animal’ housing. Before being transported by van at the MRI platform (>5min course), animals were sedated to prevent stress. Once at the MRI platform, anaesthesia was induced. Animal was then bedded in lateral recumbency and placed on a wheeled lift table and then placed on supine position onto the MRI table and maintained in this position with straps and memory foam mat and cushions to ensure stability, immobility and to prevent inadequate body pressure. Ophthalmic ointment was applied on both eyes which were maintained closed during the whole imaging procedure to avoid ocular dryness. A veterinarian was constantly present near the animal to ensure constant and conformable anaesthesia.
MRI data have been acquired on 3 Tesla VERIO Siemens systems (Erlangen, Germany), using a two flexible coil (Siemens FLEX Large 4 elements) tied around the head and three MR acquisitions were performed on each animal. MR sequences have been optimized to 1/ be performed in a time compatible with anesthesia (approximately 3h), 2/ reduce artifact (folding, truncation, etc.) and 3/ optimize SNR. Regarding these specifications the following sequences were used: - for brain morphometry investigations, a three dimensional T1w MPRAGE acquired in the coronal plane was used with the following parameters: Echo Time/Repetition Time=2.67 ms/2500 ms, Flip Angle=12°, Inversion Time=900 ms, Number of Excitation=3, Partial Fourier=1, Slice Thickness=1 mm, Slice Number=208, Field of View=256x256 mm, matrix=256x256, final resolution 1mm3. - To investigate the brain microstructure, we developed a diffusion weighted MRI protocol based on a two dimensional T2w spin-echo sequence acquired in the axial plane, over 3 different shells optimized for neurite orientation dispersion and density imaging (NODDI) model using the following parameters: Shell 1: b=300 s/mm2, 6 directions; Shell 2: b=700 s/mm2, 30 directions, and Shell 3: b=2000 s/mm2, 64 directions. The fixed parameters are Echo Time/Repetition Time=109ms/11.5s, Flip Angle=90°, Number of Excitation=1, Partial Fourier=0.75, Slice Thickness=2.4 mm, Slice Number=57, Field of View =256x256 mm, matrix=128x128, final resolution 2x2x2.4 mm3, one b=0 per shell). Sequences have been acquired in different reading phases (left-right and right-left) for distortion corrections. - To investigate brain functioning a T2w spin-echo-planar imaging (SE-EPI) sequence acquired in the axial plane and left/right reading phase was used with the following parameters: Echo Time/Repetition Time=24 ms/3.97 s, Flip Angle=90°, Number of Excitation=1, Partial Fourier=1, Slice Thickness=3.3 mm, Slice Number=40, Field of View=220x220 mm, matrix=110x110, final resolution 2x2x3.3 mm3, Number of repetition=250. Additionally, two similar sequences in different reading phases (left-right and right-left) of ten volumes each were acquired for distortion corrections.
Once the imaging procedure finished, the animal was immediately transported back to a recovery stable to wake up progressively. Hydration (3 mL of Ringer minimum), flunixine meglumine (single dose of 3.5mL, Antalzen, 50 mg/mL, laboratorios calier, Barcelona, Spain) and of acefylline heptaminol (a single dose of 25mL of Vetecardiol, Intervet MSD Santé animale, Beaucouzé, France) were systematically administered. Veterinarian and at least two technicians were constantly present near the animal to monitor its recovery until it was able to walk steadily (approximately 1h40 for recovery). Then, the animal stayed in the recovery stable at least 6 hours after he stood-up for surveillance with another animal (the mother or another foal depending on the group) to avoid social isolation stress and finally joined back their hoursing. A blood sample was taken 48h after anaesthesia and sent to the veterinary office to check the absence/presence of myositis. Only one case of myositis has been detected afterward nevertheless the animal fully recovered within a week. DICOM data from the scanner were converted to NIFTI format and organized as standardized data sets according to the Brain Imaging Data Structure (BIDS) using BIDScoin and are downloadable on Zenodo.
For voxel-based morphometry analysis (VBM), we used T1w images. Data were first denoised and signal bias corrected using Ginkgo and N4BiasFieldCorrection respectively and linearly coregistered to the TEBTA template using antsRegistrationSyNQuick. Coregistrated data were then pre-processed with SPM. Each image was segmented into probability maps of GM, WM, and CSF using the default settings in the SPM8 toolbox and the TEBTA version of GM, WM, and CSF probability maps. The transformation matrices obtained were used to normalize GM, WM and CSF probability maps obtained for each subject. GM and WM probability maps of each scanning session were normalized to our stereotaxic space using transformations matrices obtained herein and resampled. Normalized GM and WM images were used for diffeomorphic anatomical registration using exponentiated lie algebra (DARTEL) to calculate diffeormorphic flowfields. Each normalized GM image was then warped using deformation parameters calculated by the DARTEL routine and then modulated to correct the volume changes that may have occurred during the deformation step. Finally, normalized-warped-modulated GM images were spatially smoothed by convolving with a 4 mm full width at half-maximum (FWHM) isotropic Gaussian kernel to create GMC maps. To assess regional GM changes over the groups (unweaned n = 12 versus weaned n = 11), GMC maps were compared using an unpaired Student t test proposed by SPM. A brain mask was used to constrain the analysis to the brain. For each cluster, the significance of the peak voxel was set as p < 0.05 (t-score = 1.72, degree of freedom=21). The results are presented on an axial and sagittal brain slice series generated with nilearn.
Because of its practical implementation, the NODDI model has become very popular to map the tissue microstructure in vivo and ex vivo in both clinical42–45 and preclinical applications46,47. The NODDI model relies on a biophysical model that separates the diffusion of water into three diffusive compartments (intra-neurite, extraneurite and CSF)31,46, which are non-exchanging, contributing to the global diffusion attenuation. The net diffusion signal attenuation (A) corresponds to the following linear combination A resulting from a linear combination of the individual signal attenuations associated with each compartment, including: 1- the neurite compartment of water molecules trapped within axons and dendrites characterized by a volume fraction (fic), 2- the extra-neurite compartment characterized by a volume fraction (fec) and 3- the CSF compartment containing free molecules with an isotropic displacement probability characterized by a volume fraction (fiso). Hence, the net signal diffusion signal A corresponds to the following linear combination: A = fic · a.ic + fec · a.ec + fiso · a.iso We pre-processed raw multishell diffusion imaging data and mapped estimated each of the previously described fractions using Ginkgo. First multishell diffusion imaging data were corrected for magnetic susceptibility distribution (topup), motions and eddy current-induced distortions (eddy). Then mean of B0 data for each animal was calculated and resulting images were used to calculate a diffusion weighted template using by modelbuild. Resulting template was then normalized to the TEBTA T1w space using antsRegistrationSyNQuick. Then Ginkgo was used for NODDI analysis (DwiMicrostructureField) and released 4 maps: the fractional anisotropy, the intraneuritic fraction, the orientation dispersion index, and the isotropic fraction. Both linear and nonlinear transformation calculated by modelbuild and antsRegistrationSyNQuick were applied once to each contrast for spatial normalisation of the data within the TEBTA space using antsApplyTransforms. Then, images were spatially smoothed with an isotropic Gaussian kernel by convolving a 4-mm full-width at half maximum. To assess regional FA, ICF, ODI and isoF changes over the groups each contrast was compared using an unpaired Student t test proposed by SPM. A brain mask was used to constrain the analysis to brain. For each cluster, the significance of the peak voxel was set as p < 0.05 (t-score = 1.72, degree of freedom=21). The results are presented on an axial and sagittal brain slice series generated with nilearn.
rs‐fMRI data were pre‐processed as previously described17,25,27,48. Briefly, EPI images were corrected for slice timing (slicetimer), motions (antsMotionCorr) and susceptibility distribution (topup). Then mean of EPI data for each animal was calculated and resulting images were used to calculate a diffusion weighted template using by modelbuild. Resulting template was then normalized to the TEBTA T1w space using antsRegistrationSyNQuick. Then images were detrended, low- and high-pass filtered (0.01Hz – 0.08Hz) the effect of the six previously calculated motion parameters including translations and rotations, both WM and CSF and global signals, were removed from the data through linear regression using nilearn Fractional amplitude of low‐frequency fluctuations analysis The fractional amplitude of low‐frequency fluctuation (0.01–0.08 Hz) values were computed using the previously processed 4D data the fALFF function from 1000 Functional Connectomes Project was used to reveal the temporal and regional changes in GM occurring in fALFF maps. Then fALFF maps were spatially normalized to the TEBTA template using both linear and nonlinear transformation calculated previously and applied once using antsApplyTransforms. To assess regional fALFF changes over the groups each contrast was compared using an unpaired Student t test proposed by SPM. A brain mask was used to constrain the analysis to brain. For each cluster, the significance of the peak voxel was set as p < 0.05 (t-score = 1.72, degree of freedom=21). The results are presented on an axial and sagittal brain slice series generated with nilearn.