Differences in resting-state functional connectivity between depressed bipolar and major depressive disorder...

Area of Science:

• Neuroscience
• Psychiatry
• Machine Learning

Background:

• Approximately 60% of bipolar disorder (BD) patients are initially misdiagnosed as major depressive disorder (MDD), leading to suboptimal treatment.
• Early and accurate differential diagnosis between MDD and BD is essential for effective therapeutic strategies.

Purpose of the Study:

• To investigate the utility of machine learning models utilizing resting-state functional neuroimaging (rs-fMRI) features for the differential diagnosis of MDD and BD.
• To identify reliable neuroimaging biomarkers that can distinguish between MDD and BD patients during depressive episodes.

Main Methods:

• Machine learning predictive models were trained using rs-fMRI data from 62 MDD patients, 63 BD patients, and 76 healthy controls.
• Features analyzed included fractional amplitude of low-frequency fluctuations, regional homogeneity, atlas-based connectivity, seed-based connectivity, and dual regression components.
• Model performance was evaluated using permutation testing and classification accuracy.

Main Results:

• The model trained on seed-based connectivity achieved the highest classification performance, with 69.36% accuracy for BD and 63.08% for MDD.
• Seed-based connectivity also demonstrated superior performance in distinguishing MDD (78.33%) and BD (71.67%) from healthy controls.
• Connectivity patterns within the brain's reward and aversion systems were identified as critical for differentiating between MDD and BD.

Conclusions:

• Seed-based connectivity analysis using machine learning is a promising approach for the differential diagnosis of major depressive disorder and bipolar disorder.
• Distinct connectivity patterns in reward and aversion systems may represent unique neurobiological underpinnings of these disorders.
• This approach could aid in early and accurate diagnosis, leading to more personalized treatment plans.