MATLAB Toolbox Skills for Claude

As a biomedical engineering PhD student, I use MATLAB daily for medical image analysis. I created these skills because Claude often suggests Python workarounds for things MATLAB can do natively, or recommends functions that were deprecated several versions ago.

These skills give Claude accurate, toolbox-specific knowledge so it suggests code that actually works.

New to Claude Skills? Skills are knowledge packages that extend Claude's capabilities. Learn more →

See the Difference

I tested each skill by asking Claude the same question with and without the skill loaded.

How do I use MedSAM to segment a tumor from a CT volume in MATLAB?
Aspect	Without Skill	With Skill
Approach	Python bridge required	Native MATLAB
Code complexity	100+ lines across two languages	~40 lines pure MATLAB
Key function	Doesn't know it exists	medicalSegmentAnythingModel
Workflow	Temp files, subprocess calls	extractEmbeddings → segmentObjectsFromEmbeddings
3D handling	"Loop over slices" (vague)	Seed-and-propagate workflow for 3D

How do I visualize a 3D medical volume with a segmentation overlay?
Aspect	Without Skill	With Skill
Approach	Workarounds (isosurface, loops)	OverlayData parameter
Code	30+ lines	3 lines
Key syntax	Doesn't know it	volshow(V, OverlayData=L.Voxels)

---

Installation

Claude Desktop

1. Download or clone this repository
2. Zip the skill folder you want:

   zip -r matlab-medical-imaging-toolbox.zip matlab-medical-imaging-toolbox
   

3. Go to Settings → Capabilities → Skills → Customize and upload the zip
4. Toggle the skill on and start a new conversation

Claude.ai (Web)

Go to Settings → Capabilities → Skills → Customize and upload the zip file.

Claude Code

Copy the skill folder to your skills directory:

# For all your projects (personal)
cp -r matlab-medical-imaging-toolbox ~/.claude/skills/

# Or for a specific project only
cp -r matlab-medical-imaging-toolbox .claude/skills/

See the Claude Code skills documentation for more details.

---

Use with MathWorks MATLAB MCP Server

These skills work great alongside the official MATLAB MCP Core Server from MathWorks:

	What It Provides
MCP Server	Code execution, syntax checking, toolbox detection
These Skills	Toolbox-specific knowledge for accurate suggestions

See MathWorks AI resources for more tools.

---

Available Skills

Skill	What It Covers
matlab-medical-imaging-toolbox	DICOM/NIfTI I/O, MedSAM, Cellpose, radiomics, 3D visualization
matlab-image-processing-toolbox	Filtering, segmentation, morphology, watershed, regionprops
matlab-deep-learning	U-Net, semantic segmentation, custom training, transfer learning
matlab-stats-ml	Classification, regression, survival analysis, Bayesian methods
matlab-wavelet-toolbox	2D transforms, denoising, lifting schemes, shearlets

---

More Examples

How do I segment overlapping cells in a microscopy image?
Aspect	Without Skill	With Skill
Approach	Basic watershed	Production-ready watershed
Preprocessing	Simple threshold	imtophat for background correction
Edge handling	Not addressed	imclearborder for edge cases
Parameters	Generic values	Specific values with explanations

How do I create a U-Net for image segmentation in MATLAB?
Aspect	Without Skill	With Skill
Functions used	unetLayers, trainNetwork (deprecated)	Both legacy and modern functions
Modern syntax	Not mentioned	unet, trainnet (R2024b+)
Custom architecture	Not shown	Manual construction code included

How do I use shearlets for directional texture analysis?
Aspect	Without Skill	With Skill
Approach	Third-party toolbox (ShearLab 3D)	Native MATLAB Wavelet Toolbox
Setup required	Download from shearlab.org	None — built-in
Forward transform	SLsheardec2D	sheart2
External dependencies	Yes	No

---

Contributing

Found an error? Have a suggestion? Contributions are welcome.

• Report issues — Open an issue to report bugs or suggest improvements
• Submit fixes — See CONTRIBUTING.md for guidelines

All feedback is appreciated.

License

This work is licensed under a Creative Commons Attribution 4.0 International License.