A Study Of Depression Diagnostic Model Based On Multimodal Brain Imaging

Objective:Due to its various clinical symptoms and unclear pathogenesis,major depressive disorder(MDD)still has some clinical problems,such as low recognition rate,high misdiagnosis rate,and poor treatment effect.Robust and enough evidence support the idea that MDD is highly associated with brain structure and function.Hence,our study aimed to focusing on depression to explore its underpinning neural substrate using structural,functional and diffusion magnetic resonance data.Furthermore,combined with support vector machine technique,our study performed the MDD vs.BD classification model to provide objective biomarkers for disease’s diagnosis.Method:1)MDD vs.BD classification model: We performed a more effective ndividualized classification model based on support vector machine and the hole-brain “high-order functional connectivity(HOFC)” network for lassifying 73 MDD,52 BD,and 76 healthy controls,compared to conventional C-based classification models.Furthermore,we explore if clinical variables i.e.,depression severity,current medication,illness duration,and age-onset) ould affect the classification performance.2)Dynamic functional and structural connectivity of salience network in MDD and D: We measured the dynamic functional and structural connectivity of 21 DD,27 BD,and 28 healthy controls using functional MRI and diffusion urtosis imaging.Moreover,we explore the correlation between depression everity and dynamic functional and structural connectivity.3)Intrinsic gray-matter connectivity in MDD: 16 MDD and 16 healthy controls ere scanned by a 3.0T MRI scanner.We measured global and local intrinsic ray-matter connectivity based on surface-based geodesic distances.Furthermore, e performed the gray-matter connectivity– clinical variables correlation analysis.Results:1)MDD vs.BD classification model: Our study achieved a satisfactory accurary 78.53%)in MDD vs.BD differentiation.The resultant contributing features nvolved in the flexible coordination among sensory(e.g.,a MDDition,vision, nd olfaction),motor,and cognitive brain network.Despite sharing a common hronnectome pattern of affective and cognitive impairments,MDD and BD lso showed unique dynamic FC synchronization patterns.Furthermore,our tudy revealed that the age-onset of illness influenced the BD vs.MDD lassification model,with the classification performance hampered by the resence of early-onset MDD.However,no similar correlations were observed in depression severity,illness duration,and current medication.2)Dynamic functional and structural connectivity of salience network: both dynamic(but not static)functional and structural connectivity were reduced in MDD patients compared to healthy controls,particularly involvement in the left prefronto-insular pathways(L.a PFC-L.insula).The aberrant dynamic functional connectivity of L.a PFC-L.insula were significantly correlated with HAMD total scores.However,no significant effect was observed between BD patients and healthy controls.3)Intrinsic gray-matter connectivity in MDD: MDD patients showed aberrant gray-matter connectivity in left postcentral gyrus,left lateral occipital,left planum temporale,right lateral occipital,and right lingual gyrus.Furthmore,these abnormal gray-matter connectivity in MDD patients were significantly correlated with HAMD subscales and total scores.Conclusion: MDD vs.BD classification model indicated that the most contributive classification features might be implicated the dynamic coordination among sensiormotor-congnitive brain network,especially salience network,default mode network,and frontoparietal network.Internestingly,the convergent and divergent feature of MDD and BD both involved in the coordinated dynamics between salience network and frontoparietal network.Our further analysis revealed that the severity of depression is tightly associated with the prefrontal – insula pathway among the salience network.Moreover,we clarified that MDD patients had abnormal intrinsic gray-matter connectivity,suggesting abnormal intrinsic wiring cost of brain connectome.Together,our study expanded our knowledge of depression,providing potential biomarkers for its objective diagnosis.