Statistical Analysis

[sandrinebedard]

Description of statistical analysis for each image contrast (In progress)

Since we don't know exactly where the nerve root compression is, it makes more difficult agregating the data per levels

--> To check what covergae we have across participants --> check coverage for fmri --> count number of subjects for each contrast before each analysis

T2w anatomical

• SC CSA perlevel
• CSA perslice in the PAM50 image space
• Symmetry (per slice and perlevel)
• Compute R and L symmetry:
  • Register to template using only centermass-rot of step 1

sct_register_to_template -i sub-CR034_ses-baselinespinalcord_T2w.nii.gz -s sub-CR034_ses-baselinespinalcord_T2w_seg.nii.gz -ldisc sub-CR034_ses-baselinespinalcord_T2w_seg_labeled_discs_1to9.nii.gz -param step=1,type=imseg,algo=centermassrot,metric=MeanSquares,iter=10,smooth=0,gradStep=0.5,slicewise=0,smoothWarpXY=2,pca_eigenratio_th=1.6 -qc ./qc

sct_apply_transfo -i sub-CR034_ses-baselinespinalcord_T2w_seg.nii.gz -d anat2template.nii.gz -w warp_anat2template.nii.gz -o sub-CR034_ses-baselinespinalcord_T2w_seg_reg.nii.gz -x nn

• Compute dice score between L and R

• Plot CSA perslice for 2 groups

T2star

(Gray matter segmentation)

• Gray matter CSA (per level)

DWI

• ROI based: in subject space --> assess asymmetry of each subject CR vs HC, mean across all FOV

• REMEMBER: https://spinalcordtoolbox.com/user_section/tutorials/atlas-based-analysis/overcoming-the-partial-volume-effect.html

• Check sct_extract_metrics

  • Tracts:
    • Corticospinal spinal
    • Dorsal columns
    • Spino thaliamic

• Voxel-based --> in PAM50 space

  • Apply WM mask before
  • https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/TBSS/UserGuide (randomize)
  • use -n 5000, test with and without --T2

    MTS

    (T1 maps, MTsat and MTR)

• ROI based:

• Voxel-based --> in PAM50 space

  • Apply WM mask before

Functional

• Bandpass temporal filtering (see what Rob Barry does)
• nilearn check bandpass filter could be done here
• spatial smoothing --> if in template space --> no
• Replicate VV and DD connectivity
• create some masks --> centered at c5, c6, c7 --> use center labels and define a slice thickness
• pairwise --> no smoothing
• extract mean time series in ROI
• Check in subject space vs PAM50 space

[sandrinebedard]

Updated tasks:

T2w anatomical

• [x] * Create one unique metric for right left CSA (L-R)/ (L+R)
• [x] * Average dice scores at disc levels
• [x] * Average CSA and all metrics arround the 3 disc levels
• [x] * Add correlation with NDI, check DASH
• [ ] * Compare baseline vs follow up
• [ ] Merge all types of data in one unique datframe for baseline
• [ ] Run entire analysis on follow up
• [ ] preprocess sub CR36
• [ ] plot right vs left CSA for CR patients only

Functional

• [ ] * Add t-test

MT and DTI

• [ ] Analyse perlevel data