Feed

MIT engineers have adapted a light-sheet microscope to image the electrical activity of neurons across a zebrafish's entire brain, scanning the whole brain 200 times a second — once every five milliseconds. Calcium imaging, the usual proxy for neural activity, resolves activity only on the order of seconds and cannot capture single spikes, while genetically encoded voltage indicators report membrane potential directly but had previously been limited to small, localized populations. Faster camera acquisition and remote refocusing pushed volumetric imaging to rates that resolve individual neuronal impulses, revealing brain-wide activity patterns evoked by ultraviolet light; the study appears in Nature Methods.

Israel's Wearable Devices said the U.S. Patent and Trademark Office has granted it a continuation patent covering user authentication that combines gesture input with biopotential sensors recording neural signals. The technology could be applied to secure payments, user detection and authorization of sensitive actions, adding a biometric layer to gesture-controlled devices.

BrainAccess released a major update to its desktop application BrainAccess Board on July 27, 2026, featuring a rebuilt interface, improved functionality and a new AI assistant called Jena that monitors device and signal quality in real time. The company, which supplies EEG devices and software for neuroscience research and clinical use, said the update is aimed at ease of use and data management during EEG acquisition and analysis.

BrainAccess, the EEG system line from Neurotechnology, has released a MATLAB API that lets users connect its wireless dry-electrode headsets over Bluetooth, configure channels and sampling rate, start and stop data streams and pull data straight into MATLAB arrays, work that previously required Python, C or the LSL MATLAB toolbox. The toolbox installs as a single .mltbx package for desktop MATLAB R2025b or newer, mirrors the structure of the existing Python API and supports session annotations and saving directly to .mat files, so recordings can go straight into tools such as Signal Processing Toolbox and EEGLAB. BrainAccess devices, sold in MINI, MIDI, MAXI and HALO models, are widely used in neuroscience research and engineering.

French neurotech company BIOSerenity ran three months of large-scale EEG AI training experiments on Jean Zay, France's most powerful supercomputer for research, including a 300-million-parameter model trained on 110,000 hours of EEG data across 64 GPUs. It also fully trained two model families, Mercury and Neptune, in three sizes each, from about 13 million to 100 million parameters and on about 5,000 to 40,000 hours of data. Neptune improved as it scaled up while Mercury did the opposite, its smallest version performing best, a sign that brain-signal AI does not scale the way language models do; the results have not been peer-reviewed.

Paradromics has named William J. Marks chief clinical officer; he brings nearly three decades of experience in clinical neuroscience and implantable devices. The appointment comes as the company advances Connect-One, an early feasibility study of its Connexus BCI system, which is intended to restore communication for people who have lost it to spinal cord injury, stroke or ALS.

Divergence Neuro has added support for BrainBit's DragonEEG to its platform, bringing in a system with 19 EEG channels laid out on the international 10-20 system plus dedicated EMG, ECG and EOG poly channels. The integration gives clinicians, researchers, educators and developers research-grade EEG acquisition inside the Divergence ecosystem for qEEG brain mapping, neuroscience research and professional neurotechnology development, as qEEG use spreads through neurofeedback, mental health and cognitive performance work.

NeuroXess said two people with high-level paralysis played chess against each other across about 800 kilometers, between Shanghai and Nanchang in eastern China, using its "Triple-Full" (三全) brain-computer interface: one selected and placed pieces on a virtual board by thought, with the commands relayed in real time to a robot stand-in that moved the pieces at the other end, where the second participant gripped physical pieces through a brain-controlled exoskeleton glove. The company said the first participant was implanted at Huashan Hospital, Fudan University, in October 2025, moved a cursor by thought on the fifth day after surgery and reached 5.2 bits per second after 17 days of training, while the second was implanted at the First Affiliated Hospital of Nanchang University in December 2025 and was eating, drinking and writing with a brain-controlled exoskeleton one month after surgery. The outcomes come from the vendor's own account and have not been validated in peer-reviewed publication.