Development of AI for the Detection of Synchronous Burst Firing in Neural Circuit Activity

2022 Fiscal Year Final Research Report

• Project/Area Number: 21K20519
• Research Category: Grant-in-Aid for Research Activity Start-up
• Allocation Type: Multi-year Fund
• Review Section: 0403:Biomedical engineering and related fields
• Research Institution: Tohoku Institute of Technology
• Principal Investigator: Matsuda Naoki 東北工業大学, 工学部, 助教 (80909490)
• Project Period (FY): 2021-08-30 – 2023-03-31
• Keywords: 機械学習 / 脳・神経 / in vitro / 微小電極アレイ / 神経毒性 / 同期バースト発火

Outline of Final Research Achievements

This study aimed to develop a synchronization burst detection AI that can be used consistently across different experiments. Spontaneous activity of human induced pluripotent stem cell-derived neurons cultured on a microelectrode array (MEA) and responses to epileptogenic compounds were acquired. Using the obtained data, five synchronization burst detection AIs were developed, and models capable of handling inter-data variations were identified. The identified raster plot image-based learning model demonstrated the ability to detect the number of synchronization bursts with an accuracy of 99.8% and the length of synchronization bursts with an accuracy of 91.9% even with different experimental data. The synchronization burst detection method using machine learning developed in this study showed the capability to evaluate neural activity obtained from MEA consistently across different experiments.

Free Research Field

生体医工学

Academic Significance and Societal Importance of the Research Achievements

本研究で開発した機械学習を用いた同期バースト検出法は、MEAで取得した神経活動を実験間差なく評価できることが示された。国内外でMEA神経活動計測法を用いた神経毒性評価の取り組みが行われており、本研究は、当該分野の基盤かつ統一的な解析法になる位置づけである。また、本研究で行う同期バースト発火検出AIは、上述のin vitro試験のみならず、in vivoの電気活動データにも適用可能であることから、電気活動を指標とした神経科学研究全般に拡張することができ、汎用性が高い研究開発である。