On July 15, 2026, Tokai University, the National Cancer Center, and RIKEN issued a press release announcing the results of research planned and led by Professor Masamichi Takahashi of our department, conducted jointly with a research consortium including the National Cancer Center, RIKEN, and Asahikawa Medical University. Further details can be found in the Tokai University press release below.

Tokai University: https://www.tokai.ac.jp/news/detail/ai_aiidh.html

URL for the article: npj Digital Medicine: https://www.nature.com/articles/s41746-026-02695-2

In gliomas, a type of malignant brain tumor, the presence or absence of IDH gene mutations is known to be important for classifying gliomas, predicting prognosis, and determining treatment response. In this study, we developed an image diagnostic support model that uses AI to predict the presence or absence of IDH mutations from MRI images and scientifically evaluated the predictive performance of the AI model by comparing it with the image interpretation results of 18 brain tumor specialists. The results of this study showed that the AI model demonstrated superior predictive performance compared to many physicians using a brain tumor image dataset from overseas patients, but a decrease in the AI's predictive performance was observed when using a brain tumor image dataset from Japanese patients, highlighting the importance of validation tailored to the institution and patient background.

This study revealed the potential and challenges of applying AI in clinical practice. Moving forward, developing an AI system that presents not only prediction results but also uncertainties will likely lead to greater support for physicians by leveraging the strengths of AI.

If you are interested in this research, please feel free to contact Neurosurgery anytime!