neuroai

From raw neuroimaging data to state-of-the-art encoding and decoding. A Python suite for modern neuroscience research.

📦 Install

pip install neuralset          # core pipeline
pip install neuralfetch        # public dataset access
pip install neuraltrain        # deep learning models

📦 Core Packages

🧩 neuralset — Turn raw neuroimaging data into AI-ready datasets

• Data loading from local and remote sources (BIDS, custom formats)
• Event-driven processing with rich DataFrame semantics
• Multi-modal extractors for MEG, EEG, fMRI, EMG, iEEG, text, audio, video
• Composable transformations for alignment, filtering, and enrichment
• Torch-native segmentation for efficient batching

⚡ neuraltrain — Deep learning for neuroimaging

• PyTorch + Lightning for scalable model training
• Specialized architectures for MEG, EEG, fMRI, and multi-modal data
• Custom augmentations for realistic data transformations
• Domain-specific loss functions and metrics
• Multi-GPU training with distributed backends

🌐 neuralfetch — Access public brain datasets

• Interfaces to public datasets (OpenNeuro, DANDI, OSF, MOABB, ...)
• One-command downloads with automated caching and verification
• Dataset versioning for reproducible research
• Metadata enrichment and study introspection

📚 Documentation

Documentation

Infrastructure: exca documentation

🛠️ Development

Pre-commit hooks:

pre-commit install

Linting and testing:

ruff check neuralset   # Lint code
ruff format neuralset  # Format code
mypy neuralset         # Type checking
pytest neuralset       # Run tests

📂 Project Structure

• neuralset-repo/ — Data loading, transforms, extractors
• neuralfetch-repo/ — Dataset interfaces and downloads
• neuraltrain-repo/ — Deep learning models and training
• docs/ — Sphinx documentation

Third-Party Content

References to third-party content from other locations are subject to their own licenses and you may have other legal obligations or restrictions that govern your use of that content.

License

This project is licensed under the MIT License. See LICENSE for details.