The need of "biomarkers" for diagnosis and treatment for psychiatric disorders has been long recognized. In the last decade, we have seen the rapid growth of the number of neuroimaging studies that applied sophisticated machine learning algorithms and statistical methods to develop MRI-markers for major psychiatric disorders, including schizophrenia spectrum disorder (SSD), major depressive disorder (MDD), and autism spectrum disorder (ASD). While this line of research has been fruitful, it has also become clear that there exists conceptual and technical challenges that limit the impact of the developed MRI-based markers. For instance, the conceptual shift towards the dimensional approach in psychiatry raises the need for trans-diagnostic markers. Population heterogeneity and subtypes may conceptually limit the scope of a marker for a single disorder. The use of "big data" collected at multiple centers faces a number of technical challenges in order to optimally account for various instrumental and measurement biases.
This Research Topic aims to collect cutting-edge efforts to overcome the conceptual and technical limitations of the current phase (say, Phase I) of MRI-markers that are typically aimed at identifying patients of a single disorder using a single site dataset. We particularly welcome, but are not limited to, contributions that address issues arising from recent trends in psychiatry research, such as dimensional approach, population heterogeneity and subtyping, and the use of multicenter dataset. The modality of MRI could be sMRI, dMRI, resting-state fMRI, task-based fMRI, or a combination of these modalities. Disorders of interest include SSD, MDD, OCD, ASD, and ADHD, but are not strictly specified. We appreciate constructive discussion and proposal on how the present difficulties be optimally overcome and MRI marker development be transitioned into the next “Phase II”, so that we can make a substantial step forward true understanding of disease mechanisms and clinical applications.
Potential topics of interest:
• Developing trans-diagnostic MRI-markers;
• Developing MRI-markers for a major psychiatric or developmental disorder (e.g. SSD, MDD, ASD, and ADHD) using multicenter datasets;
• Developing new data harmonization methods for multicenter dataset for MRI-markers;
• Developing optimal data preprocessing protocols for sMRI, dMRI, or fMRI (e.g., choice of optimal parcellation scheme);
• Reliability and reproducibility of MRI-based measures to represent disease conditions;
• Application of new unsupervised learning algorithm for MRI-markers;
• Advanced supervised learning algorithms for single/multi-class classification, regression, and more;
• Review of the previous progress and new challenges for MRI-markers in psychiatry;
• Applications of MRI-markers for developing new intervention methods;
• Techniques in other modalities (such as M/EEG and fNIRS) applicable to the MRI-marker development.
The need of "biomarkers" for diagnosis and treatment for psychiatric disorders has been long recognized. In the last decade, we have seen the rapid growth of the number of neuroimaging studies that applied sophisticated machine learning algorithms and statistical methods to develop MRI-markers for major psychiatric disorders, including schizophrenia spectrum disorder (SSD), major depressive disorder (MDD), and autism spectrum disorder (ASD). While this line of research has been fruitful, it has also become clear that there exists conceptual and technical challenges that limit the impact of the developed MRI-based markers. For instance, the conceptual shift towards the dimensional approach in psychiatry raises the need for trans-diagnostic markers. Population heterogeneity and subtypes may conceptually limit the scope of a marker for a single disorder. The use of "big data" collected at multiple centers faces a number of technical challenges in order to optimally account for various instrumental and measurement biases.
This Research Topic aims to collect cutting-edge efforts to overcome the conceptual and technical limitations of the current phase (say, Phase I) of MRI-markers that are typically aimed at identifying patients of a single disorder using a single site dataset. We particularly welcome, but are not limited to, contributions that address issues arising from recent trends in psychiatry research, such as dimensional approach, population heterogeneity and subtyping, and the use of multicenter dataset. The modality of MRI could be sMRI, dMRI, resting-state fMRI, task-based fMRI, or a combination of these modalities. Disorders of interest include SSD, MDD, OCD, ASD, and ADHD, but are not strictly specified. We appreciate constructive discussion and proposal on how the present difficulties be optimally overcome and MRI marker development be transitioned into the next “Phase II”, so that we can make a substantial step forward true understanding of disease mechanisms and clinical applications.
Potential topics of interest:
• Developing trans-diagnostic MRI-markers;
• Developing MRI-markers for a major psychiatric or developmental disorder (e.g. SSD, MDD, ASD, and ADHD) using multicenter datasets;
• Developing new data harmonization methods for multicenter dataset for MRI-markers;
• Developing optimal data preprocessing protocols for sMRI, dMRI, or fMRI (e.g., choice of optimal parcellation scheme);
• Reliability and reproducibility of MRI-based measures to represent disease conditions;
• Application of new unsupervised learning algorithm for MRI-markers;
• Advanced supervised learning algorithms for single/multi-class classification, regression, and more;
• Review of the previous progress and new challenges for MRI-markers in psychiatry;
• Applications of MRI-markers for developing new intervention methods;
• Techniques in other modalities (such as M/EEG and fNIRS) applicable to the MRI-marker development.