Brain Co-Processors: Integrating Brain Implants With Local and Distributed Computing Resources
Abstract number :
1.086
Submission category :
2. Translational Research / 2B. Devices, Technologies, Stem Cells
Year :
2018
Submission ID :
500658
Source :
www.aesnet.org
Presentation date :
12/1/2018 6:00:00 PM
Published date :
Nov 5, 2018, 18:00 PM
Authors :
Vaclav Kremen; Benjamin H. Brinkmann; Inyong Kim, Mayo Clinic; Hari Guragain, Mayo Clinic; Mona Nasseri, Mayo Clinic; Tal Pal Attia, Mayo Clinic; Abigail Magee, Mayo Clinic; Su-Youne Chang, Mayo Clinic; Jeffrey Herron, Medtronic; Tom Adamski, Medtronic; S
Rationale: Electrical brain stimulation is an effective therapy for people with drug resistant epilepsy. Current limitations for electrical stimulation include: rare seizure free outcomes, inaccurate patient reported seizure diaries, limited analytics and stimulation flexibility for adaptive stimulation paradigms. Here we address these limitations with a novel epilepsy therapeutics platform that integrates a brain implant with local and distributed computing resources to provide real-time intracranial EEG (iEEG) telemetry, automated seizure detection and diaries, brain-state tracking, seizure forecasting, and adaptive stimulation. Methods: We implanted twelve canines with the Medtronic Inc. investigational Summit RC+S system (bilateral hippocampus & anterior nucleus of thalamus or cortical electrodes) and developed an Epilepsy Personal Assistant Device (EPAD). The EPAD is a mobile computational device that integrates the RC+S implant device with local and distributed computing resources. Low-complexity embedded algorithms can activate short-latency responsive stimulation. The EPAD provides a local computational node with bi-directional connectivity to the RC+S and cloud computing resources for large-scale data management and analytics for tracking brain state, automated seizure detection, diaries, seizure forecasting, and adaptive stimulation. Results: Twelve canines underwent monitoring using the system (average 350 days) with an average of 96% of iEEG data wirelessly telemetered and received. Two of the dogs are client owned pets, and live with the implanted device in real-world conditions. We demonstrated real-time seizure detection (average accuracy 97%, sensitivity 100%), real-time seizure forecasting (90%sensitivity, 8% time in warning) and tracking evoked responses to classify wake and sleep for brain-state adaptive brain stimulation. Conclusions: The Epilepsy Management System integrates an implanted device with local and distributed computing resources to create a bi-directional interface between brain and computers. The system provides a comprehensive platform for accurate seizure diaries, brain state tracking, and adaptive neural stimulation. Funding: This research was supported by National Institutes of Health (UH2-NS95495 & R01-NS092882), LQ1605 from the National Program of Sustainability II (MEYS CR, Czech Republic), and institutional resources from Mayo Clinic, Rochester MN USA, and Czech Technical University in Prague, Czech Republic. Conflict of Interest: Jeffrey Herron, Tom Adamski, Vince Vasoli, Elizabeth Fehrmann, Tom Chouinard, and Timothy Denison are employees of Medtronic, Minneapolis, MN USA. Brian Litt is a co-founder of Blackfynn, Inc., a provider of Cloud resources for this project. These companies contributed technical expertise but did not influence the results or content of the work.