dicom-mri — Claude Code Skill

A Claude Code skill that analyzes medical DICOM imaging studies (MRI, CT, X-ray) end-to-end — from raw files on a disc or drive, to rendered images, to written reports in clinical or plain language.

What it does

1. Discovers DICOM data from a mounted CD/DVD, USB drive, or local directory
2. Extracts patient and study metadata across all series
3. Renders every series as PNG slice-grid images so Claude can visually inspect them
4. Analyzes the imaging systematically — rotator cuff, labrum, bone marrow, tendons, cartilage, etc.
5. Generates structured reports in two formats:
   • Clinical report — for the patient's doctor, uses medical terminology
   • Plain-language report — for the patient and family, no jargon
6. Optionally generates treatment recommendations with phased recovery timelines

Install

# Clone the repo
git clone https://github.com/yamz8/dicom-mri-skill.git

# Copy the skill to your Claude skills directory
cp -r dicom-mri-skill ~/.claude/skills/dicom-mri

Or install directly in Claude Code:

/install-skill https://github.com/yamz8/dicom-mri-skill

Dependencies

# Arch / Omarchy
omarchy-pkg-add python-pydicom python-numpy python-matplotlib

# Debian / Ubuntu
sudo apt install python3-pydicom python3-numpy python3-matplotlib

# pip
pip3 install pydicom numpy matplotlib

Usage

Once installed, just tell Claude about your scan:

"I have an MRI CD in my disc reader, can you analyze it?" "My doctor gave me a CD after my knee MRI. Can you tell me what's on it in plain language?" "Analyze the DICOM files at /path/to/scan and generate a report"

Claude will handle the rest.

Bundled scripts

Script	Purpose
scripts/check_deps.py	Verify required Python packages are installed
scripts/extract_metadata.py	Print a structured summary of all series metadata
scripts/render_dicom.py	Render series to PNG overview images

Run scripts standalone

# Check dependencies
python3 scripts/check_deps.py

# View study metadata
python3 scripts/extract_metadata.py /path/to/DICOM

# Render all series to PNGs
python3 scripts/render_dicom.py /path/to/DICOM ./output

# Render only series 201, overview only
python3 scripts/render_dicom.py /path/to/DICOM ./output --series 201 --overview-only

Example output

output/
├── s201_overview.png       # PDW-SPAIR coronal — 12 representative slices
├── s201_all.png            # PDW-SPAIR coronal — all 35 slices
├── s301_overview.png       # STIR coronal
├── s501_overview.png       # T1W axial
├── s601_overview.png       # T2W-SPAIR axial
├── report_clinical.md      # structured clinical report
├── report_plain.md         # plain-language report for patient/family
└── treatment.md            # treatment recommendations (if requested)

Disclaimer

All reports generated by this skill are produced by an AI system and are not a substitute for interpretation by a licensed radiologist. Always consult a qualified healthcare professional before making any medical decisions.

License

MIT