Software tutorial presents an open-source Python platform for multimodal fNIRS and optical tomography, highlighting reproducible and machine learning-ready neuroimaging workflows.
Key Points
To introduce Cedalion, a standardized open-source Python framework designed to unify model-based and machine-learning-driven analyses of multimodal fNIRS and diffuse optical tomography data.
Integrated forward modeling, photogrammetric coregistration, motion correction, general linear model analysis, and diffuse optical tomography reconstruction within the standard Python scientific stack.
Engineered compatibility with SNIRF and BIDS data standards alongside interfaces for scikit-learn and PyTorch to support multimodal fusion with electroencephalography, magnetoencephalography, and physiological streams.
Constructed seven fully executable, containerized Jupyter tutorial notebooks with continuous integration testing and automated citation tracking.
Unified fragmented analytical pipelines into a reproducible, cloud-executable architecture for both laboratory and real-world wearable optical neuroimaging.
Enabled end-to-end execution of complex workflows spanning signal quality assessment, data augmentation, multimodal physiological analysis, and 3D tomographic image reconstruction.