A study used functional magnetic resonance imaging (fMRI) to distinguish bipolar disorder (BD) from major depressive disorder (MDD) using amygdala-based functional connectivity (FC) scans of 92 patients with MDD and 48 patients with BD. Over two years, a machine learning classifier was developed to differentiate BD from unipolar depression (UD), achieving 81% accuracy. The classifier identified 10 brain regions in the cortico-limbic neural circuit essential for this differentiation. Both patients with BD and transformed BD (tBD) exhibited similar FC patterns in this circuit, distinct from UD, with FC values not linked to the severity of mood symptoms. These patterns might serve as characteristic traits of BD. While the study presents significant findings for early diagnosis and prediction of affective disorders, limitations include a small sample size and potential medication effects, necessitating further research.
Reference: Jiang X, Cao B, Li C, et al. Identifying misdiagnosed bipolar disorder using support vector machine: feature selection based on fMRI of follow-up confirmed affective disorders. Transl Psychiatry. 2024 Jan 8;14(1):9. doi: 10.1038/s41398-023-02703-z. PMID: 38191549; PMCID: PMC10774279.
