Dendrite: Open-source Python app for real-time BCI research and development

Real-time brain-computer interface research just got a major open-source boost with Dendrite, a Python application from researchers Niko Kroflic and Jan Babič. Dendrite unifies signal acquisition, decoder training, and live inference into a single ready-to-run system that remains fully modifiable. It records multiple physiological streams simultaneously at their native rates and executes several processing modes concurrently against those streams. A decoder can start from a previously trained model or be fitted mid-session while the pipeline keeps running—and the same recordings can feed offline training within the same application. Every recording, decoder, and training run is tracked in a database, with each decoder storing its configuration and source recordings for full traceability from deployment back to creation.

A key design choice keeps the experimental paradigm external: it runs as an independent program in any language that communicates with Dendrite over the network, rather than being embedded inside the runtime. This modularity ensures researchers can leverage existing tools. The team validated Dendrite end-to-end on both in-house and public BCI datasets, successfully training and updating decoders online while the pipeline executed in real time. The code is available under GPL-3.0, making it a foundation for reproducible, open-source biomedical computing—important for labs that need reliable, documented workflows for BCI development and evaluation.