报告题目：Toward an AI-based diagnostic-aid of epileptic electroencephalogram
Toshihisa Tanaka（田中聪久） received the Ph.D. degrees from the Tokyo Institute of Technology in 2002. From 2000 to 2002, he was a JSPS Research Fellow. From October 2002 to March 2004, he was a Research Scientist at RIKEN Brain Science Institute. In April 2004, he joined Department of Electrical and Electronic Engineering, the Tokyo University of Agriculture and Technology, where he is currently a Professor.
His research interests include a broad area of signal processing and machine learning including brain and biomedical signal processing, brain-machine interfaces and adaptive systems. He is a co-editor of Signal Processing Techniques for Knowledge Extraction and Information Fusion (Springer), 2008 and a leading co-editor of Signal Processing and Machine Learning for Brain-Machine Interfaces (IET, UK), 2018.
He served as an associate editor and a guest editor of special issues in journals including Neurocomputing and IEICE Transactions on Fundamentals and Computational Intelligence and Neuroscience (Hindawi). Currently he serves as an associate editor of IEEE Transactions on Neural Networks and Learning Systems, Applied Sciences (MDPI), Advances in Data Science and Adaptive Analysis (World Scientific), and Neural Networks (Elsevier). Since 2020, he has been a Distinguished Lecturer of APSIPA. He is a senior member of IEEE, and a member of IEICE, APSIPA, Japan Epilepsy Society, and Society for Neuroscience.
Epilepsy is a chronic disorder that causes unprovoked, recurrent seizures. To diagnose epilepsy, the most common test is to measure electroencephalogram (EEG). Brain abnormalities appears as abnormal signal patterns in the recorded EEG. However, the recording typically lasts hours, sometimes days. This yields a heavy burden for clinical specialists, and the number of specialists is relatively small for the number of patients. I would like to introduce our project on development of the platform for AI-based diagnostic-aid to help specialists to interpret EEG signals. One of the topics in the project is to detect characteristic spikes, which are often observed in the EEG of epileptic patients in order to diagnose the disorder. I would like to introduce a fully data-driven method that automatically determines EEG frequency bands of interest. The raw signal is fed into a convolutional layer to detect suitable frequency bands, followed by a feedforward convolutional neural network (CNN) model or recurrent neural network (RNN) models for epileptic spike and non-spike classification. Fitting data of multiple patients, annotated by an epilepsy specialist, resulted in a convolutional layer with a frequency characteristic similar to bandpass filters. This result strongly justifies limiting the bandwidth of a signal, as done in previous studies.