Moumgiakmas Seraphim S, Vrochidou Eleni, Papakostas George A
MLV Research Group, Department of Informatics, Democritus University of Thrace, 65404 Kavala, Greece.
Artif Intell Med. 2025 Oct;168:103219. doi: 10.1016/j.artmed.2025.103219. Epub 2025 Jul 19.
The human brain is the most intricate organ, comprising trillions of synaptic connections and governing every thought, feeling, and action. However, abnormalities in its structure or dysfunction in neural connections can often underpin the development of mental disorders. Mental health conditions affect nearly 1 in 8 people globally, creating a significant challenge for healthcare systems to manage. Advances in neuroimaging and artificial intelligence (AI) hold the potential to transform mental health diagnosis and treatment by enabling the timely detection of these disorders. Radiomics, a technique that extracts quantitative features, has emerged as a promising approach for improving diagnostic accuracy and predicting treatment response. This review explores the current status of radiomics-based applications derived from neuroimaging and AI in addressing various mental disorders categorized under the fifth edition of Diagnostic and Statistical Manual of Mental Disorders (DSM). These include bipolar and anxiety disorders, depressive and neurodevelopmental disorders, schizophrenia spectrum and other psychosis, Post-traumatic stress disorder (PTSD) and Internet Gaming Disorder. The findings highlight the critical role of radiomic features and identify the brain regions associated with each disorder, alongside the tools, algorithms, and methodologies used. While the review also discusses limitations and challenges in radiomics research, it underscores the potential of radiomics and AI to identify significant biomarkers for the precise diagnosis of mental health conditions, as well as to enhance precision in treatment response. The potential of this technology could offer new approaches for the diagnosis and personalized treatment of mental disorders, ultimately improving the well-being of millions people worldwide.
人类大脑是最复杂的器官，由数万亿个突触连接组成，支配着每一个思想、情感和行动。然而，其结构异常或神经连接功能障碍往往是精神障碍发展的基础。全球近八分之一的人受到心理健康问题的影响，这给医疗系统的管理带来了巨大挑战。神经影像学和人工智能（AI）的进展有望通过及时检测这些疾病来改变心理健康诊断和治疗。放射组学是一种提取定量特征的技术，已成为提高诊断准确性和预测治疗反应的一种有前途的方法。本综述探讨了基于放射组学的应用从神经影像学和人工智能在解决《精神疾病诊断与统计手册》（DSM）第五版分类的各种精神障碍方面的现状。这些包括双相情感障碍和焦虑症、抑郁症和神经发育障碍、精神分裂症谱系及其他精神病性障碍、创伤后应激障碍（PTSD）和网络游戏障碍。研究结果突出了放射组学特征的关键作用，并确定了与每种疾病相关的脑区，以及所使用的工具、算法和方法。虽然综述还讨论了放射组学研究的局限性和挑战，但它强调了放射组学和人工智能在识别用于精确诊断心理健康状况的重要生物标志物以及提高治疗反应精准度方面的潜力。这项技术的潜力可以为精神障碍的诊断和个性化治疗提供新方法，最终改善全球数百万人的福祉。
