Brain multi-contrast, multi-atlas segmentation of diffusion tensor imaging and ensemble learning automatically diagnose late-life depression

Abstract

Abstract We aimed to develop a machine learning model for diagnostic classification in late-life major depression based on an advanced whole brain white matter segmentation framework. Twenty six late-life depression and 12 never depressed individuals aged > 55 years, matched for age, MMSE, and education underwent diffusion tensor magnetic resonance imaging and multi-contrast, multi-atlas segmentation in MRIcloud. Fractional anisotropy volume, mean fractional anisotropy, trace, axial and radial diffusivity extracted from 146 white matter parcels were used to train and test the AdaBoost classifier using 12-fold cross validation. Performance was evaluated using accuracy, balanced accuracy, precision, and recall, F1-score and area under the receiver operator characteristic curve. Statistical significance of the classifier was assessed using standard label permutation and area under the receiver operator characteristic curve scores’ comparison. The classifier achieved a balanced accuracy, of 71% and an area under the receiver operator characteristic curve of 0.81 by trace, and a balanced accuracy of 70% and an area under the receiver operator characteristic curve, of 0.80, by radial diffusivity, in limbic, cortico-basal ganglia-thalamo-cortical loop, brainstem, external and internal capsules, callosal and cerebellar structures. Both indices shared important structures for classification, while fornix was the most important structure for classification by both indices. The classifier proved statistically significant, as area under the receiver operator characteristic curve scores after permutation were lower than those with the actual data, by trace (p = 0.022) and radial diffusivity (p = 0.024). The results encourage further investigation of the implemented methods for computer aided-diagnostics and anatomically-informed therapeutics.