Proposal of an integrated monitoring system for neonatal seizures based on informations of EEG, Video and Vital signs.

2017 Fiscal Year Final Research Report

• Project/Area Number: 15K09737
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Embryonic/Neonatal medicine
• Research Institution: 独立行政法人国立病院機構岡山医療センター(臨床研究部)
• Principal Investigator: Takeuchi Akihito 独立行政法人国立病院機構岡山医療センター(臨床研究部), 独立行政法人国立病院機構 岡山医療センター(臨床研究部), 新生児科医師 (40731386)
• Co-Investigator(Kenkyū-buntansha): 曽 智 広島大学, 工学研究科, 助教 (80724351)
• Co-Investigator(Renkei-kenkyūsha): TSUJI Toshio 広島大学, 大学院工学研究科, 教授 (90179995) ; SHIMATANI Koji 県立広島大学, 保健福祉学部, 教授 (00433384) ; KOBAYASHI Katsuhiro 岡山大学, 大学院医歯薬学総合研究科, 教授 (60273984)
• Research Collaborator: NAKAMURA Makoto ; KAGEYAMA Misao ; HYODO Yuki
• Project Period (FY): 2015-04-01 – 2018-03-31
• Keywords: 新生児発作 / 動画像解析 / モニタリングシステム / 体表脈波センサ

Outline of Final Research Achievements

At first, we analyzed neonatal seizures, however, it was difficult to extract of movement due to background images. Instead of neonatal seizures, we analyzed infantile seizures (epileptic spasms) as the second best. After feature extraction of video images of infantile seizure and analysis of electroencephalogram (EEG), we calculated 2 indices of limb movements (motion variation, correlation in the movement of 4 limbs) and 2 indices of EEG signals (variation of the EEG amplitude, amplitude of fast wave). To establish the seizure detection system, neural network was trained using these 4 indices. This system could detect infantile seizures with high (more than 90%) accuracy, sensitivity and specificity. We also tried to develop a non-invasive monitoring sensor. We could separate the respiration and heart beats signals from body surface pulse signals by signal processing. We established basic technology for development of a new seizure monitoring systems for infants.

Free Research Field

胎児新生児学