According to researchers from BIFOLD/TU Berlin and Boston University, Cedalion brings a broad set of optical neuroimaging analysis tools into one open-source Python framework, centered on functional near-infrared spectroscopy (fNIRS) and diffuse optical tomography (DOT). The project addresses a familiar problem in multimodal research: analysis workflows are often assembled from fragmented tools and software environments, making methods harder to reproduce, compare, and extend. For neuroimaging teams working across MRI, EEG, and optical measurements, the significance is less about replacing one modality with another than about making multimodal experimentation easier to structure.
From raw light measurements to reconstructed images
fNIRS estimates brain activity through optical changes associated with cortical blood flow. Unlike MRI, it can be used with wearable equipment, during movement, and outside the laboratory. DOT builds on the same principle while seeking improved spatial and depth information. These properties make optical methods useful when researchers need to observe brain function in more naturalistic settings, although the resulting data still require a substantial chain of modeling, quality control, artifact correction, and reconstruction.
Cedalion is designed to place that chain within a single Python environment. The framework includes tools for simulating light propagation through tissue, estimating optode positions on the head with photogrammetry, assessing signal quality, correcting artifacts, modeling multimodal time series with general linear models, reconstructing DOT images, and applying data-driven machine-learning methods.
That breadth matters because the contrast seen in an optical measurement is not yet a direct map of neural computation. It is the endpoint of light interacting with tissue, sensors being positioned on the scalp, physiological signals entering the recording, and algorithms translating those measurements into interpretable quantities. Keeping these stages connected can make it easier for researchers to inspect how a preprocessing decision shapes the later image or statistical result.
A Python foundation for reproducible workflows
The project builds on the established MATLAB tools Homer2/3 and AtlasViewer from Boston University, carrying their methodological experience into the Python ecosystem. Its tutorial was published in Neurophotonics on 13 August 2026 and is accompanied by seven executable Jupyter notebooks. The notebooks are intended to run in the cloud without local setup, allowing neuroscientists, engineers, and data scientists to explore the framework with basic Python knowledge and without prior fNIRS experience.
This shift allows a workflow to be shared as more than a description of software steps. A complete analysis can be distributed alongside a publication and executed in the same environment, rather than reconstructed manually from separate packages. Cedalion also connects with widely used machine-learning libraries, which creates a direct route from physically and physiologically informed preprocessing to pattern extraction.
For clinical and translational researchers, that connection should be treated as infrastructure rather than as evidence of immediate diagnostic improvement. Machine-learning performance will still depend on acquisition quality, artifact handling, cohort design, validation, and the stability of the underlying measurements. A unified framework can reduce friction, but it does not remove the biological and statistical uncertainty carried by optical data.
What neuroimaging teams should watch
The immediate practical question is whether Cedalion can fit into a laboratory’s existing trajectory: the team’s current MATLAB workflows, Python tooling, multimodal acquisition protocols, and expectations for reproducibility. The open-source, community-driven design is promising for method development, but adoption will depend on documentation, interoperability, maintenance, and whether established analyses can be reproduced consistently across environments.
A community workshop is scheduled for 24–25 August 2026 in Berlin and online as part of the MoBI Conference. That timing gives researchers an opportunity to assess the framework through executable examples rather than through a feature list alone.
For MRI-focused groups, Cedalion is most relevant where optical recordings are being combined with other modalities or used to study brain function during movement and in real-world contexts. Consider the implications carefully: the value of a common framework may emerge gradually, through more transparent longitudinal workflows and easier comparison of methods, rather than through a single new image or algorithm.