Classification of Major Depressive Disorder Using Vertex-Wise Brain Sulcal Depth, Curvature, and Thickness with a Deep and a Shallow Learning Model

Abstract

Major depressive disorder (MDD) is a complex psychiatric disorder that affects the lives of hundreds of millions of individuals around the globe. Even today, researchers debate if morphological alterations in the brain are linked to MDD, likely due to the heterogeneity of this disorder. The application of deep learning tools to neuroimaging data, capable of capturing complex non-linear patterns, has the potential to provide diagnostic and predictive biomarkers for MDD. However, previous attempts to demarcate MDD patients and healthy controls (HC) based on segmented cortical features via linear machine learning approaches have reported low accuracies. In this study, we used globally representative data from the ENIGMA-MDD working group containing 7,012 participants from 30 sites (N=2,772 MDD and N=4,240 HC), which allows a comprehensive analysis with generalizable results. Based on the hypothesis that integration of vertex-wise cortical features can improve classification performance, we evaluated the classification of a DenseNet and a Support Vector Machine (SVM), with the expectation that the former would outperform the latter. As we analyzed a multi-site sample, we additionally applied the ComBat harmonization tool to remove potential nuisance effects of site. We found that both classifiers exhibited close to chance performance (balanced accuracy DenseNet: 51%; SVM: 53%), when estimated on unseen sites. Slightly higher classification performance (balanced accuracy DenseNet: 58%; SVM: 55%) was found when the cross-validation folds contained subjects from all sites, indicating site effect. In conclusion, the integration of vertex-wise morphometric features and the use of the non-linear classifier did not lead to the differentiability between MDD and HC. Our results support the notion that MDD classification on this combination of features and classifiers is unfeasible. Future studies are needed to determine whether more sophisticated integration of information from other MRI modalities such as fMRI and DWI will lead to a higher performance in this diagnostic task.

Figure 1:. Proposed conceptualization levels and implementation of classification procedure.

Left: Higher classification performance in MDD vs HC classification task can be achieved by implementing deep ML models, such as DenseNet, in comparison to a shallow ML model, for example, SVM. Furthermore, the analysis of integrated morphometric features can provide a more detailed description of cortical organization than separated features, leading to better differentiability of MDD from HC. The application of ComBat may improve the generalizability of results as site-related differences are removed. Right: Cortical sulcal depth, curvature, and thickness are first projected into the 2D grid and then transformed into 2D images using OMT projection. We split the data into 10 CV folds according to age and sex (Splitting by Age/Sex) and according to the site belonging (Splitting by Site). After the residualization step, where the age and sex effect are regressed out linearly, we train and test SVM and DenseNet on the diagnosis classification.

Figure 2:. MDD vs HC classification performance of SVM and DenseNet applied to vertex-wise cortical features.

Balanced accuracy for both classification models when trained on all features integrated with and without ComBat harmonization for both splitting strategies and when trained on single features. Error bars represent standard deviation.