Add codespell support with configuration and fixes

.codespellrc

@@ -0,0 +1,31 @@
+[codespell]
+# Ref: https://github.com/codespell-project/codespell#using-a-config-file
+skip = .git,.git-meta,.gitignore,.gitattributes,*.pdf,*.PDF,*.svg,package-lock.json,vendor,*.css,*.gii,.codespellrc,*.html,*.rtf
+check-hidden = true
+# ignore-regex =
+ignore-words-list =
+  # Proper surnames: E. Reynolds Losin, Jorge Sepulcre, Claus Svarer (neuroscientists)
+  losin,
+  sepulcre,
+  claus,
+  # Journal abbreviations: Hum Brain Mapp (Human Brain Mapping), Nat Commun (Nature Communications)
+  mapp,
+  commun,
+  # Neuroimaging acronyms: TE=Echo Time, FWE=Family-Wise Error, BU=author initials, ANS=Autonomic Nervous System
+  te,
+  fwe,
+  bu,
+  ans,
+  # MATLAB variable/algorithm identifiers: indx=index, som=self-other-map, ttest=t-test function, als=alternating-least-squares
+  indx,
+  som,
+  ttest,
+  als,
+  # Brain region abbreviation: PreS=Presubiculum
+  pres,
+  # Meta-analysis coding category in data files (column name / cell value)
+  anyother,
+  # Dutch surname: Peter De Weerd (neuroscientist, appears in references.bib)
+  weerd,
+  # MATLAB command 'whos' (lists workspace variables; appears in debug session transcript)
+  whos

.github/workflows/codespell.yml

@@ -0,0 +1,23 @@
+# Codespell configuration is within .codespellrc
+---
+name: Codespell
+
+on:
+  push:
+    branches: [master]
+  pull_request:
+    branches: [master]
+
+permissions:
+  contents: read
+
+jobs:
+  codespell:
+    name: Check for spelling errors
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v6
+      - name: Codespell
+        uses: codespell-project/actions-codespell@8f01853be192eb0f849a5c7d721450e7a467c579  # v2.2

Atlases_and_parcellations/2013_Shen_Constable_NIMG_268_parcellation/Shen_MNI152NLin2009cAsym_create_atlas_object.m

@@ -100,7 +100,7 @@
 
 %% write - this writes only the label image
 
-% Save: if creating from probabilty images only
+% Save: if creating from probability images only
 %
 % if dosave
 %     

Atlases_and_parcellations/2013_Shen_Constable_NIMG_268_parcellation/Shen_MNI152NLin6Asym_create_atlas_object.m

@@ -100,7 +100,7 @@
 
 %% write - this writes only the label image
 
-% Save: if creating from probabilty images only
+% Save: if creating from probability images only
 %
 % if dosave
 %     

Atlases_and_parcellations/2013_Shen_Constable_NIMG_268_parcellation/Shen_create_atlas_object.m

@@ -100,7 +100,7 @@
 
 %% write - this writes only the label image
 
-% Save: if creating from probabilty images only
+% Save: if creating from probability images only
 %
 % if dosave
 %     

Atlases_and_parcellations/2014_Keuken_7T_subcortex/Keuken_create_atlas_object.m

@@ -104,7 +104,7 @@
 
 %% write - this writes only the label image
 
-% Save: if creating from probabilty images only
+% Save: if creating from probability images only
 %
 % if dosave
 %     

Atlases_and_parcellations/2016_Fan_Brainnetome_r273_parcellation/Brainnetome_create_atlas_object.m

@@ -91,7 +91,7 @@
 
 %% write - this writes only the label image
 
-% Save: if creating from probabilty images only
+% Save: if creating from probability images only
 %
 % if dosave
 %     

Atlases_and_parcellations/2016_Glasser_Nature_HumanConnectomeParcellation/METHODS.md

@@ -102,7 +102,7 @@ https://github.com/Washington-University/HCPpipelines/blob/master/global/templat
 https://github.com/Washington-University/HCPpipelines/blob/master/global/templates/standard_mesh_atlases/fs_R/fs_R-to-fs_LR_fsaverage.R_LR.spherical_std.164k_fs_R.surf.gii
 onto the subject specific registered spherical surface ([lr]h.sphere.reg in the freesurfer output 
 directory). This was then projected into the native subject surface space using the inverse transforms from
-the fmriprep outputed freesurfer directories  and then transformed from native space into 
+the fmriprep outputted freesurfer directories  and then transformed from native space into 
 MNI152NLin2009cAsym and MNI152NLin6Asym spaces. The results for the left and right hemisphere
 concatenated across subjects for each space, study and hemisphere.
 

Atlases_and_parcellations/2016_Glasser_Nature_HumanConnectomeParcellation/README.md

@@ -25,13 +25,13 @@ version for improvements now available.
 
 ## Probabilities
 
-The atlas is probablistic. The original atlas is defined on a surface, but this surface is folded in different 
+The atlas is probabilistic. The original atlas is defined on a surface, but this surface is folded in different 
 ways for different people. Misalignment of gyri and sucli results in parcels projecting into different locations
 for different people. The probabilities indicate the likelihood an individual's parcel will project to the 
 designated location. For instance, V1 and V2 probability maps are shown below, and although one is always more
 probable than another there are also regions where identity at the group level is ambiguous.
 
-![Examples of probablistic labels](images/probability_maps_vis_ctx.png)
+![Examples of probabilistic labels](images/probability_maps_vis_ctx.png)
 
 You can threshold the map at different probability values to obtain more or less eroded versions of the atlas 
 using the atlas object's threshold method. This offers a more rigorous dilation/erosion method than was available 
@@ -42,7 +42,7 @@ for the older glasser atlas. Some example thresholds are shown below.
 Probabilities were computed first based on alignments in each of three studies (BMRK5, PainGen, SpaceTop) separately
 and then averaged. Different studies have different acquisition parameters, which can result in systematic
 differences in resulting alignments. Averaging over more studies would be better. In the interest of facilitating
-futher updates to this dataset these three studies have had their subject specific MNI space projections uploaded
+further updates to this dataset these three studies have had their subject specific MNI space projections uploaded
 to figshare. If you want to recompute the probability maps/parcel boundaries you can follow the approach in
 the src subfolder here for a new study and combine it with these maps for an incremental improvement. Maps are here:
 
@@ -76,7 +76,7 @@ from a study previously analyzed and reported in surface space.
 
 To my knowledge Glasser et al. have not published their subject specific region classifiers (as of 10/24/23), 
 nor a multimodal alignment template, so subject specific parcel identification is not yet possible outside the
-Van Essen lab. However, subject specific parcellations may be avialable from the HCP dataset, and might provide 
+Van Essen lab. However, subject specific parcellations may be available from the HCP dataset, and might provide 
 an avenue towards an improved version of this atlas if someone were to obtain appropriate surface to volume
 mappings from the HCP study. Coalsen et al. 2018 PNAS achieve precise this kind of surface to volume mapping 
 that takes variability of the subject specific surface parcellations into account and compares it to a projection 
@@ -91,7 +91,7 @@ were also computed. By enchaining the inverse transform to surface space with th
 space we can get a fsaverage surface to MNI space projection for each subject. The Glasser surface parcellation
 was transformed with this enchained set of transforms for each of 241 unrelated participants in paingen, 88
 participants from BMRK5 and 112 participants from SpaceTop. All transformations were computed by running fmriprep 
-20.2.3 (BMRK5, PainGen) or 21.0.2 (SpaceTop) with recon-all enabled on these participants, so this probablistic 
+20.2.3 (BMRK5, PainGen) or 21.0.2 (SpaceTop) with recon-all enabled on these participants, so this probabilistic 
 atlas is ideally calibrated for use with fmriprep aligned volumes. Resulting parcellations were uploaded to 
 figshare
 
@@ -143,7 +143,7 @@ precision, it's a matter of an imprecise fit to a mean template.
 What we offer here instead is the mean of many precise fits to individual templates, which is clearly better in this 
 regard. While different individuals may have different ventral visual cortical foldings, none of them have have visual 
 cortex in the cerebellum. Finally, the old atlas is necessarily non-probablistic, which limits its versatility. Note that 
-in the figure below underlay is MNI152NLin2009cAsym, which is nearly indistiguishable from MNI152NLin2009aAsym, the 
+in the figure below underlay is MNI152NLin2009cAsym, which is nearly indistinguishable from MNI152NLin2009aAsym, the 
 asymmetric version of the template the old glasser parcellation was projected to. The fmriprep20 registration fusion 
 image it cycles with is thresholded at 20% probability, which is a default threshold used throughout much of this repo.
 
@@ -186,7 +186,7 @@ Alternatively we can look at the difference between studies, since we have three
 greatest agreement, with practically all dice coefficients greater than 0.7. This is fortunate, since it means
 between study variability is small relative to between subject variability, and it means we're closer to the number
 of degrees of freedom we need for stable estimates of parcel boundaries (Using a single study, so no study related 
-error, Wu et al. claim you need 300 participants for boundaries to fully stabalize).
+error, Wu et al. claim you need 300 participants for boundaries to fully stabilize).
 
 <img alt="paingen vs. bmrk5 to MNi152NLin2009cAsym" src="diagnostics/dice_hist_interstudy_glassers_MNI152NLin2009cAsym.png" width="500" />
 

Atlases_and_parcellations/2016_Glasser_Nature_HumanConnectomeParcellation/old/GlasserHCP_create_atlas_object.m

@@ -111,7 +111,7 @@
 
 %% write - this writes only the label image
 
-% Save: if creating from probabilty images only
+% Save: if creating from probability images only
 %
 % if dosave
 %     

Atlases_and_parcellations/2016_Glasser_Nature_HumanConnectomeParcellation/old/README.md

@@ -1,4 +1,4 @@
 The code in this folder is deprecated. It's based on a dubious projection from surface space to volumetric space
 that's based on nearest neighbor interpolation without seemingly any realignments performed. A registration
 fusion approach was used to generate a more credible alignment and is available in the parent directory. This is
-left here for backwards compatability.
+left here for backwards compatibility.

Atlases_and_parcellations/2018_Iglesias_thalamic_reconstruction/README.md

@@ -1,15 +1,15 @@
 ## Overview
 
-This is a probablistic atlas meant to correspond to freesurfer's thalamic 
+This is a probabilistic atlas meant to correspond to freesurfer's thalamic 
 subnuclear parcellations, a joint histologic and in vivo parcellation of the
 thalamus into 23 unique regions per hemisphere. It is less granular than 
-Morel, but has an open usage license, is probablistic and is likely to be 
+Morel, but has an open usage license, is probabilistic and is likely to be 
 more accurately segmented. That notwithstanding, the biological accuracy of 
 the labeling hasn't been sufficiently validated in my opinion (nor has any 
 other MNI space thalamic atlas), so treat it primarily as an ontology rather 
 than something with any kind of granular biological validity.
 
-The probablistic labels are derived from labels produced by application of an
+The probabilistic labels are derived from labels produced by application of an
 automated bayesian segmentation algorithm to 618 participants from 3 studies.
 The most recent versions of the algorithm is robust when given DWI and T1 
 data, so if individual parcellations are needed you can obtain them easily.
@@ -21,17 +21,17 @@ as well.
 
 ## Methods
 
-The probablistic labels are not directly derived from the original Iglesias
+The probabilistic labels are not directly derived from the original Iglesias
 et al. (2018) Neuroimage parcellation that is used internally by freesurfer.
-Those probablistic labels aren't readily available. They're stored in an
-esoteric format interally by freesurfer, and based on personal correspondence
+Those probabilistic labels aren't readily available. They're stored in an
+esoteric format internally by freesurfer, and based on personal correspondence
 with Iglesias there is no readily available nifti version of the
-probablistic labels available for distribution. There is a version available
+probabilistic labels available for distribution. There is a version available
 that's packaged with LeadDBS, but the relationship between this version and
-the original probablistic labels isn't clear. Instead, what I (BP) have done
+the original probabilistic labels isn't clear. Instead, what I (BP) have done
 is to fit the freesurfer segmentation to 278 HCP participants, 76 spacetop
 participants and 264 paingen participants and use those subject specific labels
-to determine probablistic labels for the population.
+to determine probabilistic labels for the population.
 
 Atlas labels were generated by running freesurfer's Thalamic segmentation on
 T1 and DWI data from 278 unrelated HCP participants, 76 spacetop participants 
@@ -116,17 +116,17 @@ all labels would be 1). Signals obtained from individualized parcellations shoul
 therefore be more homogeneous across participants than signals obtained from the 
 group parcels. We therefore take task evoked responses from SpaceTop (Pain, Vicarious 
 and Cognitive) and estimate the mean parcel task evoked response for each in each 
-thalamic region in each of the 76 particpants segmented using both individual parcellations 
+thalamic region in each of the 76 participants segmented using both individual parcellations 
 and a group parcellation. We then evaluate the effect of parcellation type and the
 interaction between task x parcellation type for each region. We would expect to see 
 larger regional effects with the individualized parcellation due 
 to constructive amplification while we expect lower regional effects in the group 
-parcellation due to destructive interference of signals from heterogenous sources 
+parcellation due to destructive interference of signals from heterogeneous sources 
 ("blurring"). We also expect task differences between regions to be greater in the 
 individualized parcellation (significant interactions) for the same reason.
 
 First level GLM contrasts were estimated using task evoked responses taken
-with respect to baseline, controling for 24 motion parameters and CSF, and using a 
+with respect to baseline, controlling for 24 motion parameters and CSF, and using a 
 canonical HRF for convolution. Data was unsmoothed. Of the 76 participants with individual 
 parcellations, 73 of them had task contrasts available (the other three were presumably 
 missing BOLD data, I ran all that I could).
@@ -136,7 +136,7 @@ parcel using either an individualized parcellation or a group parcellation. Subj
 treated as random effectively turning this into a within subjects repeated measures 
 analysis, although this isn't clearly illustrated in the figures below, so take note now.
 Task contrasts were coded pain = [1/2, -1/3] vic = [-1/2, -1/3], cog = [0, 2/3] which 
-yeilded two planned contrasts: pain_v_vic and pain_vic_v_cog. A parcellation variable was 
+yielded two planned contrasts: pain_v_vic and pain_vic_v_cog. A parcellation variable was 
 also coded as [1/2, -1/2] for the individual vs. group parcellation (resp). In Wilkinson 
 notation the model took the form,
 
@@ -151,7 +151,7 @@ it's important to note.
 
 Models were fit separately for each region using restricted expectation maximum 
 likelihood (ReML) and full covariance estimation. Satterthwaites correction was used to 
-estimate degrees of freedom, and in most cases yeilded n-1 degrees of freedom, but due 
+estimate degrees of freedom, and in most cases yielded n-1 degrees of freedom, but due 
 to some regions missing in some individual parcellations the df are also somewhat variable.
 This is precisely what we would expect for a repeated measures analysis.
 
@@ -161,7 +161,7 @@ with respect to baseline. Of the 76 participants with individual parcellations,
 BOLD data, I ran all that I could).
 
 The results are very mixed. With 46 regions our Holm-Sidak corrected p-value 
-theshold is p < 0.0011 for alpha=0.05.
+threshold is p < 0.0011 for alpha=0.05.
 
 Regions with lower evoked response across tasks using individualized parcellations (unexpected):
 - central_medial_(left)<br />
@@ -254,7 +254,7 @@ The same as the above only shown in FSL with a T2 image juxtaposed along side it
 
 The morel atlas is the most detailed histological atlas available in MNI space.
 Its major limitation is that it is based on 6 participants (or maybe 3 and
-mirror imaged, unclear), is not probablistic, and lacks an open license. The 
+mirror imaged, unclear), is not probabilistic, and lacks an open license. The 
 Iglesias atlas is based on 12 participants (or maybe 6 and mirror imaged) 
 augmented by 55 in vivo scans that are used to calibrate probabilistic maps. 
 In our case these are further mediated by projection to 278 individual's 
@@ -351,5 +351,5 @@ Right Thalamus (axial)
 
 ## References
 
-* Iglesias JE, Insausti R, Lerma-Usabiaga G, Bocchetta M, Van Leemput K, Greve DN, van der Kouwe A, Fischl B, Caballero-Gaudes C, Paz-Alonso PM. (2018). A probablistic atlas of the human thalamuc nuclei combining ex vivo MRI and histology. Neuroimage, 314-326, 183.
+* Iglesias JE, Insausti R, Lerma-Usabiaga G, Bocchetta M, Van Leemput K, Greve DN, van der Kouwe A, Fischl B, Caballero-Gaudes C, Paz-Alonso PM. (2018). A probabilistic atlas of the human thalamuc nuclei combining ex vivo MRI and histology. Neuroimage, 314-326, 183.
 * Tregidgo HFJ, Soskic S, Althonayan J, Maffei C, Van Leemput K, Golland P, Insausti R, Lerma-Usabiaga G, Caballero-Gaudes C, Paz-Alonso PM, Yendiki A, Alexander DC, Bocchetta M, Rohrer JD, Iglesias JE. (2023). Accurate Bayesian segmentation of thalamic nuclei using diffusion MRI and an improved histological atlas. Neuroimage,  274, 120129.

Atlases_and_parcellations/2018_Iglesias_thalamic_reconstruction/compare_with_leadDBS_version.m

@@ -64,7 +64,7 @@
 thisAtlas.probability_maps = [];
 thisLeadsAtlas = leadsAtlas.select_atlas_subset(thisAtlas.labels,'exact').threshold(0.2);
 thisLeadsAtlas.probability_maps = [];
-for orientation = {'saggital','coronal','axial'}
+for orientation = {'sagittal','coronal','axial'}
     %%
     o2 = thisAtlas.montage('transvalue',0.5,'regioncenters',orientation{1});
     for i = 1:num_regions(thisAtlas)
@@ -92,7 +92,7 @@
 thisAtlas.probability_maps = [];
 thisLeadsAtlas = leadsAtlas.select_atlas_subset(thisAtlas.labels,'exact').threshold(0.2);
 thisLeadsAtlas.probability_maps = [];
-for orientation = {'saggital','coronal','axial'}
+for orientation = {'sagittal','coronal','axial'}
     %%
     o2 = thisAtlas.montage('transvalue',0.5,'regioncenters',orientation{1});
     for i = 1:num_regions(thisAtlas)

Atlases_and_parcellations/2018_Iglesias_thalamic_reconstruction/compare_with_morel.m

@@ -52,7 +52,7 @@
 thisAtlas = [newAtlas1{:}];
 thisMorelAtlas = [newAtlas2{:}];
 
-for orientation = {'saggital','coronal','axial'}
+for orientation = {'sagittal','coronal','axial'}
     %%
     o2 = thisAtlas.montage('transvalue',0.5,'regioncenters',orientation{1});
     for i = 1:num_regions(thisAtlas)
@@ -84,7 +84,7 @@
 thisAtlas = [newAtlas1{:}];
 thisMorelAtlas = [newAtlas2{:}];
 
-for orientation = {'saggital','coronal','axial'}
+for orientation = {'sagittal','coronal','axial'}
     %%
     o2 = thisAtlas.montage('transvalue',0.5,'regioncenters',orientation{1});
     for i = 1:num_regions(thisAtlas)

Atlases_and_parcellations/2018_Iglesias_thalamic_reconstruction/iglesias_MNI152NLin2009cAsym_create_atlas.m

@@ -4,7 +4,7 @@
 
 space_description = 'MNI152NLin2009cAsym';
 atlas_name = sprintf('iglesias_HCP278_ST76_PG264_%s', space_description);
-references = char({'Iglesias JE, Insausti R, Lerma-Usabiaga G, Bocchetta M, Van Leemput K, Greve DN, van der Kouwe A, Fischl B, Caballero-Gaudes C, Paz-Alonso PM. (2018). A probablistic atlas of the human thalamuc nuclei combining ex vivo MRI and histology. Neuroimage, 314-326, 183.'; ...
+references = char({'Iglesias JE, Insausti R, Lerma-Usabiaga G, Bocchetta M, Van Leemput K, Greve DN, van der Kouwe A, Fischl B, Caballero-Gaudes C, Paz-Alonso PM. (2018). A probabilistic atlas of the human thalamuc nuclei combining ex vivo MRI and histology. Neuroimage, 314-326, 183.'; ...
              'Tregidgo HFJ, Soskic S, Althonayan J, Maffei C, Van Leemput K, Golland P, Insausti R, Lerma-Usabiaga G, Caballero-Gaudes C, Paz-Alonso PM, Yendiki A, Alexander DC, Bocchetta M, Rohrer JD, Iglesias JE. (2023). Accurate Bayesian segmentation of thalamic nuclei using diffusion MRI and an improved histological atlas. Neuroimage,  274, 120129.'});
 
 dosave = true;