Investigating structural-functional brain covariation in bipolar disorder using a multimodal fusion approach

Area of Science:

• Neuroscience
• Psychiatry
• Medical Imaging

Background:

• The neurobiological basis of bipolar disorder (BD) is not fully understood due to inconsistent findings across imaging modalities.
• Existing research often analyzes neuroimaging data in isolation, limiting comprehensive insights into BD pathophysiology.

Purpose of the Study:

• To investigate the neurobiological underpinnings of bipolar disorder using a multimodal fusion algorithm.
• To integrate regional homogeneity (ReHo), gray matter volume (GMV), and fractional anisotropy (FA) from MRI scans to identify group-discriminative patterns.

Main Methods:

• A data-driven multimodal fusion algorithm was applied to MRI data from 125 BD patients and 113 healthy controls (HCs).
• Joint independent components (jICs) were generated by fusing ReHo, GMV, and FA data to differentiate between BD patients and HCs.
• Covariant patterns across the three modalities were analyzed.

Main Results:

• The fusion algorithm successfully differentiated BD patients from HCs across the three modalities.
• BD patients exhibited an inverse functional pattern in the default mode network (DMN) and sensorimotor network (SMN), with increased ReHo in the DMN and decreased ReHo in the SMN compared to HCs.
• This inverse pattern was mirrored in GMV, with increased volume in the DMN and decreased volume in the SMN. The precuneus showed functional hyperactivation and decreased structural volume.

Conclusions:

• Multimodal neuroimaging fusion offers a more comprehensive understanding of bipolar disorder pathophysiology.
• The identified inverse functional and structural patterns in specific brain networks provide valuable insights for advancing BD diagnosis and treatment.
• The precuneus plays a significant role in cognitive function within the context of BD.