Synthetic Data
Cedalion provides tools for generating synthetic fNIRS data for two purposes:
Algorithm development and testing — synthetic HRFs and motion artefacts can be added to real or noise-only recordings to benchmark preprocessing and detection algorithms under controlled conditions where the ground truth is known.
Machine learning benchmarks — the BimodalToyDataSimulation in
cedalion.sim.datasets generates paired synthetic fNIRS+EEG datasets with
controllable signal-to-noise ratio, frequency band, inter-modality time lag, and
mixing matrix structure. It can be used to create reproducible benchmark
experiments for multimodal decomposition methods.
Functions for generating synthetic artifacts in fNIRS data. |
|
Functions for generating synthetic hemodynamic response functions. |
