Brain Implant Reads Neural Conversations to Treat Depression and PTSD

Deep brain stimulation has long promised relief for patients with severe psychiatric disorders, but current systems suffer from a critical limitation: they monitor only single brain regions, missing the complex communication between neural networks that underlie conditions like depression, PTSD, and anxiety. Now, researchers at EPFL have developed a solution that could transform treatment approaches.

The team, led by Mahsa Shoaran at EPFL's Integrated Neurotechnologies Lab, has created LogSync—a system-on-chip that tracks communication patterns across multiple brain regions while consuming less than 200 microwatts of power. That's significantly lower than typical implantable neural interfaces, making long-term brain monitoring feasible without frequent battery replacements or bulky power systems.

From Single Sites to Neural Networks

Traditional deep brain stimulation delivers fixed electrical pulses to specific brain regions, an approach that has shown promise for movement disorders but limited success in psychiatry. The problem, according to the EPFL team, is that psychiatric conditions often stem from disrupted communication between brain areas rather than dysfunction in a single location.

LogSync addresses this by extracting biomarkers that describe both brain wave strength and synchrony between different regions. Using machine learning algorithms integrated directly into the chip's hardware, the device can distinguish between cognitive and emotional states from network-level brain activity in real time. This enables adaptive stimulation that responds to a patient's actual brain state rather than delivering predetermined pulses.

The innovation relies on logarithmic computations that simplify the complex mathematical operations required for brain-network analysis, allowing the chip to perform sophisticated processing while consuming minimal power. The team validated their approach using human brain recordings from psychiatric experiments involving cognitive engagement and emotional arousal tasks, as well as in vivo studies in rats conducted at the University of Minnesota.

Clinical Applications and Development

The research, published in the IEEE Journal of Solid-State Circuits, targets severe psychiatric illnesses including major depression, PTSD, obsessive-compulsive disorder, and mood and anxiety disorders. The work is supported by a CHF 6.6 million grant from the US National Institute for Mental Health.

For healthcare providers, the technology represents a potential shift from single-region to network-based neuromodulation, enabling more personalized therapy that adapts to real-time brain states. For patients with treatment-resistant conditions, network-level adaptive stimulation could improve outcomes by targeting the disrupted communication patterns underlying their symptoms, potentially reducing side effects while improving therapeutic precision.

The ultra-low power consumption and integrated hardware-algorithm design also establish a benchmark for implantable neural interfaces beyond psychiatry. Applications could extend to movement disorders, non-invasive stimulation systems, and focused-ultrasound interfaces.

Shoaran's team is advancing related technology through Infera Neuro, an EPFL spin-off she co-founded. The researchers also presented self-supervised learning methods for neural activity analysis at the International Conference on Machine Learning, suggesting continued innovation in how AI interprets brain signals.

As psychiatric disorders continue to challenge conventional treatments, LogSync's ability to monitor the brain's internal communication—rather than isolated regions—offers a more nuanced approach to understanding and treating these complex conditions.

Sources

1. Low-power implantable neural chip tracks communication across brain regions — News-Medical.net