2023 Volume 34 Issue 1 Pages 19-23
There is an expectation that the findings from human brain magnetic resonance imaging (MRI) for psychiatric disorders could differentiate diagnostic categories and predict future prognosis. However, obtained differences between patients and healthy controls have a considerable overlaps and there has been no developed biological markers from MRI studies. Recently, machine learning and deep learning methodologies which use a lot of variables have been popular, but to utilize the techniques, we need to properly harmonize the MRI data from multiple sites and procedures. We continue to explore for the limitation of the MRI data from multi sites, and build machine learning classifiers which could help to use in clinical settings and to elucidate pathophysiology of psychiatric disorders. For the upcoming large‐sample brain MRI studies, we established a novel MRI protocol with high‐resolution multi‐modal images, and developed a traveling subject method to harmonize the data. In the future, collaborative studies are expected to enable a high‐quality multi‐site MRI harmonization and to promote international collaboration for standardized diagnosis of neuro‐psychiatric diseases with neuroimaging.