Transcranial motor evoked potential (Tc-MEP) is used in many institutions to monitor spinal cord function intraoperatively. In this study, we examined how the amplitude of Tc-MEP changes when the train repetition frequency (TRF) of the multi-train stimulation (MTS) is set at 5 Hz and 10 Hz. The results showed that the amplitude was higher in the tibialis anterior and abductor pollicis brevis muscles of the lower limb at 5 Hz compared to 10 Hz (p<0.01). The amplitude was higher in the brachioradialis and short abductor pollicis brevis muscles of the upper limb at 10 Hz compared to 5 Hz (p<0.01). It was considered that changing the TRF settings in the Tc-MEP MTS is useful as a means to obtain stable waveform derivation in cases where monitoring is difficult due to low amplitude.

With advances in AI technologies such as deep learning and big data, research and development aimed at introducing AI into medical examinations and treatments have been progressing. In some cases, these technologies have obtained regulatory certification or approval as medical devices following clinical trials. This trend is no exception in the field of clinical neurophysiology. Among the projects in which the authors are involved, an implantable brainmachine interface (BMI) has reached the stage of initiating clinical trials, a fully automated analysis system for magnetoencephalography in epilepsy has completed development, and a wearable BMI device designed to promote rehabilitation have obtained medical device certification. In light of these developments, the Japanese Society of Clinical Neurophysiology has established an AI and Big Data Working Group (WG), and its subordinate body focusing on development and implementation: a Development and Application Sub-Working Group (SWG). The Development and Application SWG will support society members in promoting the use of AI and big data technologies within the society. Specifically, it plans to: (1) propose and conduct society-led research on development and application; (2) provide expertise on intellectual property strategy, regulatory (PMDA) strategy, and reimbursement in collaboration with the Education and Outreach SWG; (3) serve as a liaison for interdisciplinary and industry collaboration in cooperation with the Dataset SWG; and (4) promote dissemination and application of research outcomes in collaboration with the Rules and Guidelines SWG. Although these plans are currently at the proposal stage, we intend to move forward toward their implementation.

Electroencephalography (EEG) and magnetoencephalography (MEG) examinations in patients with epilepsy require substantial expertise and experience on the part of the neurophysiologists. Because those waveforms must be visually inspected one by one, the process is time-consuming. In MEG, dipole analysis further increases the workload: for each epileptiform waveform, the examiner must determine the optimal time point for analysis between the onset and the peak of the spike. This decision must be made individually for every waveform, resulting in considerable time and effort. Artificial intelligence (AI) -based big data analysis has the potential to automate such labor-intensive and expertise-dependent interpretation and analysis, thereby improving the efficiency of clinical practice. However, AI systems trained on data from a specific device or institution often exhibit decreased performance when applied to data from different devices or institutions. This issue of limited generalizability, which is not typically observed in human interpretation, represents a major challenge. To overcome this problem, AI requires strategies distinct from those used by humans, such as collecting large-scale datasets from diverse devices and institutions and converting them into standardized formats for unified analysis. Moreover, although human readers can efficiently improve their interpretive skills by learning from pitfalls and exceptional cases once they have acquired basic competence, AI systems do not necessarily benefit in the same way. In some cases, additional training on specific exceptions may even degrade previously acquired performance (catastrophic forgetting), highlighting current limitations of AI methodologies. Nevertheless, given the rapid progress in AI technology, these challenges are likely to be mitigated in the near future. Although current systems remain imperfect, the automation of interpretation and analysis using AI-driven big data approaches holds significant value, not only for enhancing clinical efficiency but also for promoting standardization in EEG and MEG interpretation. In this article, we discuss these issues with a particular focus on the complete automation of dipole analysis in MEG for epilepsy.

In intraoperative neurophysiological monitoring (IONM), spontaneous electromyographic activity (sEMG) is a valuable modality that enables real-time detection of intraoperative neural irritation, providing immediate feedback to the surgeon. By using sEMG complementarily with other IONM modalities, such as transcranial motor evoked potentials and somatosensory evoked potentials, sEMG significantly contributes to enhancing the safety of spinal surgery.