While recent advances in MRI technology have enabled high‐precision brain imaging, challenges remain in data quality consistency across different MRI systems and research reproducibility. Brain/MINDS Beyond addresses these issues through a large‐scale MRI study across 18 Japanese institutions, encompassing over 7,000 participants (healthy controls and neuropsychiatric patients) . The program introduced the Harmonized Protocol (HARP) for standardization across different MRI manufacturers and models. Using traveling subjects to assess inter‐scanner variations, we achieved high‐quality multimodal MRI data within a 30‐minute protocol, including structural (0.8mm isotropic) , functional (2.4mm isotropic, 10min) , and diffusion MRI (1.7mm isotropic) . The study established standardized preprocessing methods based on the HCP Pipeline and developed various analytical approaches for investigating brain functional and microstructural architecture. The resulting high‐precision database and analytical framework provide a robust foundation for understanding brain organization and neuropsychiatric pathophysiology, with promising applications in clinical diagnosis and prognosis prediction.

View full abstract

Brain structural alterations in psychiatric disorders are being progressively elucidated through the analysis of large‐scale brain magnetic resonance imaging (MRI) data from multiple sites. However, biases arising from differences in scanning procedures and sample characteristics between MRI sites have compromised the reliability of the analyses and consistency of interpretation. The authors here intend to propose a method integrating harmonization based on multi‐site imaging of the same subject (traveling subject ; TS) and the statistical method ComBat‐GAM to assess disorder‐specific structural brain changes in psychiatric disorders. This method suggests the possibility of data harmonization without burdensome TS scans while reducing measurement bias. Furthermore, by estimating normative lifespan trajectories using nonlinear regression with large‐scale healthy control data, we can highlight disease‐common and ‐specific variation, enabling highly sensitive analysis beyond conventional meta‐analysis and flexible analysis, including the addition of new data. Based on the harmonized data, it is expected that the characteristics of brain structural changes for each psychiatric disorder will be clarified in the future, contributing to the elucidation of the pathophysiology of the disorders and the improvement of diagnostic accuracy.

View full abstract

Resting‐state functional connectivity (rsFC) methods have been used in basic and clinical research as a way to visualize and quantify functional brain networks. Although psychiatric disease biomarker research and development for diagnostic and therapeutic targeting and subtyping is underway, establishing a practical biomarker remains a challenge. In this study, to comprehensively and quantitatively assess the impact of various factors related to biomarker development on FC, we have analyzed a large travel subject data set from the Brain Mind/Beyonds Project (BMB, 2018‐2023) and a multi‐disorder data set measured in SRPBS (2012‐2018) , consisting of total 2,100 runs of the functional magnetic resonance imaging (fMRI) data during a 10‐minute eye‐open resting sate experiment. The results showed that the two main non‐disease factors, intra‐individual inter‐trial and inter‐individual variability, were the two most important factors, with disease variability being as large as the measurement factors, scanner and measurement protocol‐to‐protocol variability. On the other hand, there were a small number of connectvity in which disease variability was the largest. When the brain networks with the highest variability for each factor were examined, it was observed that different networks were affected with overlap between the factors. These results suggest that for reliable biomarker development, it is necessary to select the connectivity with large disease variability and small other factors, and to develop experimental and analytical methods to reduce intra‐ and inter‐individual variability.

View full abstract

Diffusion MRI is a technique that enables the evaluation of white matter microstructure by utilizing water molecule diffusion, and its utility as an early diagnostic and prognostic marker in various neuropsychiatric disorders has been suggested. However, conventional white matter microstructure models are known to be unable to accurately estimate fiber structures when multiple fiber bundles exist within a single voxel. Recently developed fibre orientation distribution (FOD) has gained attention as a model that can distinguish multiple fiber directions within a single voxel using spherical deconvolution methods, overcoming this limitation. This paper provides an overview of the basic principles and limitations of conventional microstructure models using diffusion MRI, and the development of FOD. We also introduce Fixel‐based analysis, a white matter structure evaluation method based on FOD, and discuss its utility in neuropsychiatric disorders. Looking ahead, we anticipate that the application of diffusion MRI quantitative techniques to large‐scale cohort data of thousands of cases, such as Brain/MINDS Beyond, will accelerate our understanding of the pathophysiology of neuropsychiatric disorders and the development of novel biomarkers.

View full abstract

With the movements toward the legalization of casinos in Japan, gambling disorder is drawing attention as a significant social issue. Here, this article reviews an attempt to examine the differences in decision‐making processes between individuals with gambling disorder and healthy controls through the application of reinforcement learning models to behavioral and neuroimaging data. First, I provide an overview of computational neuroscience research, particularly focusing on decision‐making and reinforcement learning. The concept of “lack of behavioral flexibility,” which is closely linked to a core symptom of gambling disorder (i. e., difficulty in stopping gambling despite adverse outcomes) , is then discussed within the context of reinforcement learning. Finally, I present our recent study in computational psychiatry, which explores the neural basis of behavioral inflexibility in individuals with gambling disorder by combining reinforcement learning and functional magnetic resonance imaging (fMRI) . No potential conflictis of interest were disclosed.

View full abstract

Depression has severe impacts, such as causing labor loss, increasing the risk of lifestyle‐related diseases, and raising suicide risk, with potential effects extending to future generations. This underscores the importance of early prediction and intervention. Advances in machine learning are expected to improve prediction accuracy, with applications expanding to classifying depression, predicting treatment response, and assessing recurrence risk. This paper reviews recent trends and challenges in machine learning research on early prediction of depression. Despite numerous studies, the accuracy of predicting new onset of depression remains low. There is also a lack of approaches that comprehensively utilize indicators related to abnormalities in emotional and reward systems‐mechanisms linked to the onset of depression‐alongside traditional demographic and psychosocial factors. Achieving high accuracy with fewer features remains a challenge. Moreover, AI tuning is often insufficient, highlighting the need for efficient optimization methods such as Bayesian optimization. Additionally, the imbalance in depression prevalence necessitates the use and optimization of PR curves and F1 scores as evaluation metrics. This paper discusses the future prospects for building practical prediction models in light of these challenges.

View full abstract

Genetic and epidemiological studies suggest that autoimmunity may underlie schizophrenia. A key component of autoimmune pathology is the presence of autoantibodies. Recently, novel synaptic autoantibodies have been identified in patients with autoimmune encephalitis, leading to the concept of autoimmune psychosis-an acute psychosis associated with positive autoantibodies. Based on these backgrounds, we hypothesized the existence of unidentified synaptic autoantibodies contributing to the pathology of schizophrenia and conducted investigations to identify them. Consequently, we discovered novel autoantibodies targeting synaptic adhesion molecules NCAM1 and NRXN1. We demonstrated that when these autoantibodies were isolated from patients and introduced into the cerebrospinal fluid of mice, they disrupted the binding of NCAM1 and NRXN1 at synapses. This interference led to a reduction in synaptic density and spines, resulting in impaired cognitive function, abnormal prepulse inhibition, and reduced sociability. These findings suggest that the autoantibodies detected in schizophrenia patients could be pathogenic and represent potential therapeutic targets. Furthermore, these autoantibodies may serve as biomarkers to identify subgroups of schizophrenia patients who could benefit from such treatments.

View full abstract