Mschallenge Skill

MS Challenge Skill (Dataset-Orchestration Layer)

Overview

mschallenge-skill is the NeuroClaw orchestration skill for the Longitudinal MS Lesion Segmentation Challenge dataset.

It strictly follows the NeuroClaw hierarchical design principles:

• This skill only describes WHAT needs to be done and which tool skill to delegate to.
• It contains no implementation code or concrete commands.
• All concrete execution is delegated to existing base/tool skills via claw-shell.
• Companion scripts in scripts/ provide reference implementations for data validation, lesion analysis, and QC.

Core workflow (never bypassed):

1. Identify input MS Challenge data and target modalities.
2. Generate a numbered execution plan clearly stating WHAT needs to be done and which tool skill will handle each step.
3. Present the full plan, estimated runtime, resource requirements, and risks to the user and wait for explicit confirmation ("YES" / "execute" / "proceed").
4. On confirmation, delegate every step to the appropriate skill via claw-shell.
5. After execution, save all outputs in a clean directory structure (mschallenge_output/).

Research use only.

Quick Reference

Dataset Characteristics

• Origin: ISBI 2015 Longitudinal MS Lesion Segmentation Challenge
• Training: 5 subjects, each with 2 timepoints (longitudinal)
• Testing: 14 subjects (hidden ground truth), 4-6 timepoints each
• Modalities: T1w, T2w, FLAIR, PD (co-registered)
• Ground truth: Manual lesion segmentation masks (training only)
• Resolution: ~0.5 × 0.5 × 0.5 mm (isotropic)
• Preprocessing: Skull-stripped, co-registered to common space
• Reference: Carass et al. (2017), NeuroImage

Supported Modalities

Directory Structure (Native)

training/
├── subject01/
│   ├── time01/
│   │   ├── subject01_time01_T1.nii.gz
│   │   ├── subject01_time01_T2.nii.gz
│   │   ├── subject01_time01_FLAIR.nii.gz
│   │   ├── subject01_time01_PD.nii.gz
│   │   └── subject01_time01_lesion.nii.gz  (ground truth)
│   └── time02/
│       └── ...

BIDS Preparation

Script: scripts/validate_mschallenge.py

Validates MS Challenge directory structure and generates a compliance report.

python skills/mschallenge-skill/scripts/validate_mschallenge.py \
  --input /path/to/MSChallenge/training \
  --output /path/to/mschallenge_output/qc/validation.csv

Features:

• Directory structure validation
• Modality completeness check (T1w, T2w, FLAIR, PD)
• Ground truth mask presence verification
• Longitudinal timepoint consistency
• Missing data identification

Core Workflow (Never Bypassed)

1. Identify user target: full MS Challenge processing, lesion analysis, or validation only.
2. Generate a numbered plan with tools, outputs, runtime, storage, and risks.
3. Wait for explicit confirmation (YES / execute / proceed).
4. On confirmation, run data validation using scripts/validate_mschallenge.py.
5. Delegate to smri-skill for structural MRI processing.
6. If lesion analysis is requested, run scripts/analyze_lesions.py.
7. If longitudinal analysis is requested, run scripts/longitudinal_lesion.py.
8. If QC summary is requested, run scripts/mschallenge_qc_summary.py.
9. Save outputs into mschallenge_output/.

Modality Processing Delegation

Standard Output Layout

mschallenge_output/
├── raw/                    # Original MS Challenge files
├── validation/             # Validation reports
├── smri/                   # Structural MRI derivatives
├── lesions/                # Lesion analysis results
│   ├── lesion_stats.csv
│   └── longitudinal_change.csv
├── qc/                     # QC summaries and exclusion lists
└── logs/                   # Processing logs

Benchmark Adapter Guidance

For benchmark-style prompts, do not force the full orchestration when the task only asks for local MS Challenge data validation.

• If the task starts from MS Challenge data already present on disk and only asks for validation:
  • Skip the download stage
  • Default to the narrow path local MS Challenge discovery -> validation -> report
• In benchmark mode, do not require explicit confirmation before presenting the validation solution.

Safety and Execution Policy

• No execution before explicit plan confirmation.
• All execution must be routed via claw-shell.
• Missing dependencies must be resolved by dependency-planner before running.

Important Notes and Limitations

• MS Challenge is a longitudinal dataset; consider timepoint effects in analysis.
• Ground truth masks are only available for training subjects.
• All images are preprocessed (skull-stripped, co-registered).
• Lesion segmentation is the primary task; standard brain morphometry may be affected by lesions.
• The challenge is designed for benchmarking; results should be compared with published baselines.
• mschallenge-skill is orchestration-only; detailed preprocessing logic remains in modality skills.

When to Call This Skill

• User asks for end-to-end MS Lesion Challenge workflow.
• User asks to validate MS Challenge data structure.
• User asks for lesion volume and count analysis.
• User asks for longitudinal lesion change tracking.
• User asks for MS lesion segmentation benchmarking.

Complementary / Related Skills

• smri-skill → structural MRI preprocessing
• nibabel-skill → NIfTI I/O and mask manipulation
• brain-visualization → lesion overlay visualization
• dependency-planner → dependency resolution
• conda-env-manager → environment management
• claw-shell → command execution

Reference

• Carass et al. (2017): Longitudinal multiple sclerosis lesion segmentation: Resource and challenge. NeuroImage.
• ISBI 2015 MS Lesion Challenge: https://smart-stats-tools.org/lesion-challenge

Created At: 2026-05-06 13:31 HKT Last Updated At: 2026-05-06 13:31 HKT Author: chengwang96