Abstract:Abstract Nearly a quarter of bipolar disorder (BD) patients were misdiagnosed as major depressive disorder (MDD) patients, which cannot be corrected until mania/hypomania develops. It is important to recognize these obstacles so that the appropriate treatment can be initiated. Thus, we sought to distinguish patients with BD from MDD, especially to identify misdiagnosed BD before mania/hypomania, and further explore potential trait features that allow accurate differential diagnosis independent of state matters. Functional magnetic resonance imaging scans were performed at baseline on 92 MDD patients and 48 BD patients. The MDD patients were then followed up for more than two years. After follow-up, 23 patients transformed into BD (tBD), and 69 patients whose diagnoses remained unchanged were eligible for unipolar depression (UD). A support vector machine classifier was trained on the amygdala-based functional connectivity (FC) of 48 BD and 50 UD patients using a novel region-based feature selection. Then, the classifier was tested on the dataset, encompassing tBD and the remaining UD. It performed well for known BD and UD and can also distinguish tBD from UD with an accuracy of 81%, sensitivity of 82.6%, specificity of 79%, and AUC of 74.6%, respectively. Feature selection results revealed that ten regions within the cortico-limbic neural circuit contributed most to classification. Furthermore, in the FC comparisons among diseases, BD and tBD shared almost overlapped FC patterns in the cortico-limbic neural circuit, and both of them presented pronounced differences in most regions within the circuit compared with UD. The FC values of the most discriminating brain regions had no prominent correlations with the severity of depression, anxiety, and mania/hypomania (FDR correction). It suggests that BD possesses some trait features in the cortico-limbic neural circuit, rendering it dichotomized by the classifier based on known-diagnosis data.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to distinguish between bipolar disorder (BD) and major depressive disorder (MDD), especially to identify BD patients who were misdiagnosed as MDD before the appearance of manic or hypomanic symptoms. Through functional magnetic resonance imaging (fMRI) technology, based on the functional connectivity (FC) of the amygdala, the research uses a support vector machine (SVM) classifier for diagnostic distinction. In addition, the study also aims to explore potential feature markers that can accurately make differential diagnoses independently of disease states, thereby improving the diagnostic dilemma between BD and MDD and providing a new perspective for predicting affective disorders. ### Main Objectives: 1. **Distinguish BD and MDD**: Especially identify those BD patients who were misdiagnosed as MDD before the first appearance of manic or hypomanic symptoms. 2. **Explore potential feature markers**: Look for brain region features that can effectively distinguish BD and MDD in different disease states. ### Methods: - **Data collection**: The study included 140 patients, of which 92 were MDD patients and 48 were BD patients. MDD patients were followed up for more than two years, and finally 23 MDD patients were transformed into BD (tBD), and the diagnoses of the remaining 69 patients remained unchanged (UD). - **Imaging analysis**: Use functional magnetic resonance imaging (fMRI) to obtain baseline data, and perform pre - processing and functional connectivity analysis. - **Machine learning model**: Use a support vector machine (SVM) classifier, based on the functional connectivity data of the amygdala, to train the model to distinguish BD and UD. At the same time, use random forest (RF) for feature selection to further screen out the most discriminative brain regions. ### Results: - **Classification performance**: In the test data set, the accuracy rate of the SVM classifier is 81%, the sensitivity is 82.6%, the specificity is 79%, and the AUC value is 74.6%. - **Key brain regions**: The study found that ten brain regions, including the left postcentral gyrus, right inferior temporal gyrus, left middle frontal gyrus, right lingual gyrus, right middle frontal gyrus, left middle occipital gyrus, right middle temporal gyrus, left precentral gyrus, left inferior parietal gyrus and left inferior temporal gyrus, contribute the most to the classification. - **Differences in functional connectivity**: The functional connectivity patterns of BD and tBD in these key brain regions are similar, and there are significant differences compared with UD. These differences are mainly concentrated within the cortico - limbic neural circuit. - **Symptom correlation**: Although there is a certain correlation between the functional connectivity of some brain regions and the severity of depression, anxiety and manic/hypomanic symptoms, after FDR correction, these correlations are not significant. ### Discussion: - **Stability of feature markers**: The research results show that the functional connectivity characteristics of BD and tBD within the cortico - limbic neural circuit have relatively high stability and are not affected by short - term emotional states. This suggests that these characteristics may be the inherent properties of BD, not only determined by the current emotional state. - **Clinical application prospects**: By using these stable feature markers, more accurate diagnostic tools can be developed to help clinicians identify BD patients at an early stage, thereby avoiding misdiagnosis and inappropriate treatment. In conclusion, this study provides new methods and perspectives for the differential diagnosis of BD and MDD, which is helpful for improving the clinical management of these two affective disorders.

Identifying misdiagnosed bipolar disorder using support vector machine: feature selection based on fMRI of follow-up confirmed affective disorders

DISTINGUISHING BIPOLAR DEPRESSION FROM MAJOR DEPRESSIVE DISORDER USING FNIRS AND DEEP NEURAL NETWORK

Screening for Bipolar Disorder with the Mood Disorders Questionnaire in Patients Diagnosed As Major Depressive Disorder — the Experience in China

Abnormal functional connectivity within the reward network: a potential neuroimaging endophenotype of bipolar disorder.

Brain-gut Microbiota Multimodal Predictive Model in Patients with Bipolar Depression

Identification of bipolar disorder using a combination of multimodality magnetic resonance imaging and machine learning techniques

Early Identification of Bipolar from Unipolar Depression Before Manic Episode: Evidence from Dynamic Rfmri

Explainable machine-learning algorithms to differentiate bipolar disorder from major depressive disorder using self-reported symptoms, vital signs, and blood-based markers

Differentiation of Transformed Bipolar Disorder from Unipolar Depression by Resting-State Functional Connectivity Within Reward Circuit.

Individualized identification of first-episode bipolar disorder using machine learning and cognitive tests

Combined static and dynamic functional connectivity signatures differentiating bipolar depression from major depressive disorder

Machine learning with multiple modalities of brain magnetic resonance imaging data to identify the presence of bipolar disorder

Divergent and convergent imaging markers between bipolar and unipolar depression based on Machine Learning

Machine learning-based discrimination of unipolar depression and bipolar disorder with streamlined shortlist in adolescents of different ages

Brain functional networks and structures that categorize type 2 bipolar disorder and major depression

Innovative Neuroimaging Biomarker Distinction of Major Depressive Disorder and Bipolar Disorder through Structural Connectome Analysis and Machine Learning Models

Intelligent classification of major depressive disorder using rs-fMRI of the posterior cingulate cortex

Detection study of bipolar depression through the application of model-based algorithm in terms of clinical feature and peripheral biomarkers

Machine Learning Algorithm Performance Evaluation in Structural Magnetic Resonance Imaging-Based Classification of Pediatric Bipolar Disorders Type I Patients.

Based on white matter microstructure to early identify bipolar disorder from patients with depressive episode

Enhancing Early Diagnosis of Bipolar Disorder in Adolescents through Multimodal Neuroimaging