N. Trung Doan
Multimodal fusion of brain magnetic resonance imaging data: methodological development and application to mental disorders.
Nhat Trung Doan
Novel biomarkers for desease diagnosis
In neuroimaging studies, it is becoming more and more common that magnetic resonance imaging (MRI) data of different modalities are acquired for each subject, allowing for studying different aspects of the brain such as functional brain networks (functional MRI), anatomical connections (diffusion tensor imaging), and brain structure (structural MRI). In the literature, these modalities are often processed and interpreted separately. Currently, there is an increasing interest in joint analysis of multimodal MR images or the fusion of their features to find common variation as well as take advantage of the complementary information offered by each modality. Computational methods for multimodal fusion are highly desirable to a vast number of brain imaging studies, including studies of mental disorders. Such methods would potentially open the doors for the discovery of novel multivariate multidimensional MR-based biomarkers for disease diagnosis, disease monitoring and evaluation the effect of drug therapies.
Need for further Insights on the underlying pathophysioloygy
Schizophrenia and bipolar disorder are highly heritable mental disorders. Studies using either structural MRI or functional MRI have provided evidence on brain structural and functional alterations in these disorders. Nevertheless, the complex dynamics and interactions of the brain structure and function in psychotic disorders necessitate investigations using multimodal imaging data to gain further insights on the underlying pathophysiology.
The overall objectives of this research topic are:
- To develop and utilize multimodal analysis techniques allowing the fusion of brain MR data, for instance the joint modeling of functional and structural connectivity indices obtained from diffusion tensor imaging and functional MRI, in the context of psychosis research.
- To search for multimodal features that could lead to maximal classification accuracy between patients (schizophrenia, bipolar disorder) and healthy controls and between the disorders.
- To explore the genetic underpinning of the brain structure and function in psychosis by means of association studies using genetic data and multimodal MR features.
Among other ongoing projects, we are currently focusing on identifying MR-based cortical "fingerprint" of psychosis by means of the fusion of cortical morphology indices such as cortical thickness, arealization, local gyrification index and gray matter concentration. Preliminary results have shown strong association between diagnosis and the derived multimodal features.
We expect that multivariate analyses based on multimodal fusion techniques will yield novel insights on the brain structural and functional breakdown in psychosis, and thus bringing added value to diagnosis and interventions.