[10p-N302-1~10] AI for Measurement Science: Toward New Measurement Paradigms Enabled by the Deep Integration of Information and Measurement Sciences

4:15 PM - 4:45 PM JST(7:15 AM - 7:45 AM UTC)

[10p-N302-7] Development of AI-based waveform classifiers for real-world neural recording and manipulation

〇Seitaro Iwama¹, Junichi Ushiba¹ (1.Keio Univ.)

Keywords:

electroencephalogram,waveform classification,brain-computer interface

Scalp electroencephalography (EEG), which enables real-time and non-invasive monitoring of brain activity, is a promising signal source for real-world brain–machine interfaces (BMIs) and neurofeedback. However, scalp EEG is limited by signal attenuation and by noise arising from body movements and electrode–skin contact. In this lecture, I will focus on the sensorimotor rhythm (SMR) and introduce a waveform-classification AI pipeline that integrates individual frequency identification, 1/f correction, burst detection, and autoencoder-based noise detection. I will further discuss how brain-state estimation can be applied to feedback systems, highlighting its potential to visualize and modulate motor cortical states and thereby induce changes in motor function.