Back to the list
Congress: ECR25
Poster Number: C-24068
Type: Poster: EPOS Radiologist (scientific)
Authorblock: M. Dostál, M. Keřkovský, T. Rohan, J. Bednařík, M. Němec; Brno/CZ
Disclosures:
Marek Dostál: Nothing to disclose
Miloš Keřkovský: Nothing to disclose
Tomas Rohan: Nothing to disclose
Josef Bednařík: Nothing to disclose
Martin Němec: Nothing to disclose
Keywords: CNS, MR-Diffusion/Perfusion, Imaging sequences, Normal variants, Image verification
Methods and materials

Methods

Following ethics committee approval, all 71 neurologically healthy volunteers provided written informed consent for this prospective study. The cervical spinal cord MRI (C1/2-C7/Th1) was performed in all subjects on a 3T scanner (19-channel head-neck coil) following the protocol outlined in Table 1.

Table 1: MRI protocol. Sag – Sagital, Ax – Axial, STIR – Short time inversion recovery, IVIM – Intravoxel incoherent motion, FOV – Field of view, Acq. Voxel. – Acqusition voxel size, TR – Repetition time, TE – Echo time, TI – Inversion time, ETL – Echo train length, NEX – Number of excitation, ARC - Autocalibrating Reconstruction for Cartesian imaging, HS – HyperSense, Seq. Name – Name of sequence by vendor, Acq. Time – Acqusition time
Anatomical images were reviewed by radiologists, and subjects exhibiting spinal cord compression were excluded. Similarly, IVIM data were inspected, and slices affected by artifacts (motion, susceptibility, low SNR, etc.) were removed from further analysis.

Data processing employed the SpinalCord Toolbox (SCT) [6], DIPY [7], and FSL [8]. Sagittal T2-weighted images were automatically segmented to delineate the spinal cord (sct_deepseg_sc) and identify vertebral body levels (sct_label_vertebrae), with manual annotation of posterior disc positions (sct_label_utils). Axial anatomical images (Ax MERGE) were then segmented into gray matter (GM) and white matter (WM) (sct_deepseg_sc). IVIM data underwent motion artifact correction (sct_dmri_moco), followed by voxel-wise IVIM parameter fitting using DIPY's variable projection method (IvimModelVP). Subsequent steps included segmentation (sct_deepseg_sc), registration of axial anatomical and IVIM images to the PAM50 atlas (sct_register_multimodal), inverse warping of the atlas to individual subject space (sct_warp_template), and extraction of IVIM parameters at C2-C7 (sct_extract_metric) for statistical analysis.

Quality control was implemented throughout the processing pipeline. All SCT commands included the -qc argument, and FSL/DIPY outputs were visually inspected. Processing parameters were adjusted or manual corrections were performed as needed to address errors.

Statistical analyses were conducted at each vertebral level (C2-C7). Welch's t-test, paired t-test, and multivariable linear regression were used to assess the influence of demographic factors. Statistical analyses were performed using R and Statistica 14.
GALLERY