skills.rest
CUHK-AIM-Group
CUHK-AIM-GroupOFFICIAL
Explore all skills by this author →
WebsiteGitHub

dictlearning

Decompose resting-state fMRI into sparse network components using dictionary learning.

89|5|Updated Mar 25, 2026
Search Tags:#connectivity#neuroimaging#fmri#nilearn#dictionary learning#resting-state#network decomposition
One-click install
npx skills add https://github.com/CUHK-AIM-Group/NeuroDiscovery --skill dictlearning-cuhk-aim-group
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: dictlearning
Source: https://github.com/CUHK-AIM-Group/NeuroDiscovery/tree/main/skills/dictlearning
Command: npx skills add https://github.com/CUHK-AIM-Group/NeuroDiscovery --skill dictlearning-cuhk-aim-group

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires nilearn.

What problem does it solve? Researchers analyzing resting-state fMRI need to identify intrinsic brain networks without labeled training data, and manually configuring sparse decomposition pipelines is error-prone and hard to reproduce. ## Core Features & Use Cases - Resting-State Network Decomposition: Fits sparse dictionary learning on preprocessed rs-fMRI to extract interpretable network component maps. - Subject-Level Time Series: Produces per-subject component time series suitable for downstream connectivity, clustering, or group comparison. - Delegation Workflow: Coordinates with fmri-skill for preprocessing and nilearn-tool for concrete DictLearning fitting and component export. - Use Case: Given a cohort of preprocessed resting-state BOLD images and a group mask, fit a 20-component dictionary decomposition and export component maps, time series, and a correlation matrix for connectivity analysis. ## Quick Start Ask the agent to run DictLearning decomposition on your preprocessed resting-state fMRI list with a group mask and 20 components, after confirming preprocessing is complete.

Frequently Asked Questions about dictlearning

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run dictionary learning on resting-state fMRI data?▼

Provide a list of preprocessed resting-state BOLD images and an optional group mask, then run the nilearn-tool DictLearning reference script with your chosen number of components. Preprocessing and denoising should be completed first via fmri-skill.

What is DictLearning used for in fMRI analysis?▼

DictLearning is a non-deep-learning unsupervised method that decomposes resting-state fMRI into sparse network component maps and subject-level time series. It is used for discovering intrinsic connectivity networks rather than task activation or phenotype prediction.

Does DictLearning require preprocessed fMRI data?▼

Yes, DictLearning expects preprocessed resting-state BOLD images, optionally with confounds tables and a brain mask. If preprocessing, nuisance regression, filtering, or standard-space alignment is not done, delegate those steps to fmri-skill first.

How does the number of components affect dictionary learning results?▼

The number of components strongly controls decomposition granularity: fewer components yield coarse large-scale networks, while more components produce finer parcellations. Results are also sensitive to preprocessing quality, head motion, filtering, and masking choices.

Can DictLearning perform statistical group comparison directly?▼

No, DictLearning is unsupervised and does not provide statistical group inference on its own. Group comparisons require additional statistical analysis applied to the exported subject-level time series or connectomes after decomposition.

Related Skills

Explore More Education & Research Skills→
NeuroAIHubNeuroAIHub
269

fmriprep

Execute containerized BIDS-App fMRI preprocessing with motion correction and anatomical registration.

Official
Advanced
xjtulycxjtulyc
33

connectome-analysis

Compute functional connectivity matrices and brain network graph metrics from ROI time series.

Community
Advanced
djkloopdjkloop

critique

Evaluate interfaces for UX quality, accessibility, and AI-generated anti-patterns.

Community
Advanced
snoodleboot-iosnoodleboot-io

feature-engineering

Guides encoding, transformation, and leakage-safe feature construction for tabular machine learning.

Official
Intermediate