Lesion-Symptom Mapping Guide (Claude Skill)

Guides Claude through inferring brain–behaviour relationships from patient lesion data — segmenting and normalizing lesions, choosing between voxel-wise and multivariate mapping, and correcting for the spatial dependence that makes naive statistics wrong here.

   
Type Claude Skill
Supplier Awesome Cognitive and Neuroscience Skills (community OSS, MIT)
Availability GA — one of ~40 research skills in the collection
Pricing Free / OSS (MIT)
Capabilities Read/Write — methodology guidance; Claude writes and runs the analysis code locally
Verified works · 2026-08-10
Security cleared · 2026-08-10 — MIT, org transferred to NeuroAIHub (confirmed), old URL still redirects

How to install

  • Claude Code — plugin marketplace (installs all skills in the collection): 
    /plugin marketplace add HaoxuanLiTHUAI/awesome_cognitive_and_neuroscience_skills
/plugin install awesome-cognitive-and-neuroscience-skills@awesome-cognitive-and-neuroscience-skills

    Restart Claude Code afterwards. The skills are description-activated — there is no slash command; ask a lesion-mapping question and Claude loads the skill.

  • Claude Code — single-skill alternative. This skill declares a required dependency on the collection’s research-literacy skill, so copy both: 
    git clone https://github.com/HaoxuanLiTHUAI/awesome_cognitive_and_neuroscience_skills
cp -r awesome_cognitive_and_neuroscience_skills/skills/lesion-symptom-mapping-guide ~/.claude/skills/
cp -r awesome_cognitive_and_neuroscience_skills/skills/research-literacy ~/.claude/skills/

    (Project-scoped alternative: copy into .claude/skills/ instead. The repo’s default branch is master, not main.)

  • Underlying software — the skill is method guidance, not an analysis engine, and it pins no versions. Most of the toolchain it names is MATLAB-based (NiiStat, VLSM2, the SVR-LSM toolbox); LESYMAP is R, BCBToolkit is a GUI/Python disconnection package, and registration runs through ANTs or FSL.

What it does

Runs a five-stage protocol: research planning → lesion segmentation → registration → statistical analysis → validation.

Method selection by cohort size and question:

  • N ≥ 50 with a continuous outcome → VLSM (mass-univariate voxel-wise)
  • N ≥ 100 and distributed patterns expected → SVR-LSM (multivariate; the skill puts its own floor at N ≥ 80–100)
  • N < 50 → ROI-based or descriptive analysis only, on the grounds that voxel-wise mapping is underpowered
  • White-matter pathway questions → disconnection analysis
  • Network-level questions → lesion network mapping

Thresholds it enforces: a voxel is tested only where lesion overlap reaches ≥ 10% of the sample (or N ≥ 10 patients), with ≥ 15% recommended; multiple comparisons are corrected by permutation-based FWE with 5,000+ permutations, thresholded at the 5th percentile of the maximum-statistic distribution.

Segmentation and registration: manual tracing is treated as the gold standard, with semi-automated (LINDA, lesion_gnb) and deep-learning alternatives; registration requires cost-function masking, and enantiomorphic normalization for large lesions.

Primary use cases: stroke and aphasia cohort studies, clinical neuropsychology, mapping post-surgical or traumatic deficits.

Notes

AI-generated content — verify before use. All skills in this collection carry review_status: ai-generated, and the README states the content “has not been individually verified by human domain experts.” This skill cites Bates et al. 2003, Kimberg et al. 2007, Foulon et al. 2018, Zhang et al. 2014, Boes et al. 2015 and Sperber 2020 — check the sample-size floors and overlap thresholds against those sources. No tool version numbers are pinned upstream.

The correction choice is the load-bearing part. Lesion voxels are spatially correlated and lesion anatomy is driven by vascular territory, so the skill states that lesion maps violate the independence assumptions behind FDR and parametric corrections. Permutation FWE is the recommended default, Benjamini-Hochberg FDR an acceptable alternative, Bonferroni “too conservative; almost never detects effects”, and uncorrected results never acceptable for publication. Lesion volume is a mandatory covariate.

Related catalogued tools: fMRIPrep and Nilearn for the preprocessing and image-handling layers, FreeSurfer for anatomy, and NetNeuroTools Guide for the network metrics and null models a lesion network mapping analysis feeds into.

Sources

Installed this tool?

Share feedback — install path, OS, errors, workarounds. The form opens with this tool pre-selected and a link back to this page.