Late Life Depression Automated Detection by Means of Advanced Diffusion Tensor Imaging Segmentation and Artificial Intelligence

Although late-life depression (LLD) is a common mental health problem in seniors and linked with dementia, its diagnosis remains challenging. Early treatment is vital and depending on accurate diagnosis. Brain white matter (WM) changes revealed by Diffusion Tensor Imaging (DTI) have been identified in LLD, yet results exhibit variability. We propose a methodology for LLD automated classification based on an advanced DTI segmentation framework and artificial intelligence (AI). Twenty-six patients and 12 healthy individuals (HC) aged > 55 years, matched for age, education, and MMSE participated. Five DTI measures were calculated within 146 WM segments. A dataset was constructed by the extracted features from each subject and used to train and test the AdaBoost, Multilayer Perceptron Artificial Neural Network and Random Tree classifiers to automatically detect LLD and HC. Due to class imbalance, Synthetic Minority Over-sampling Technique was applied. Our initial results encourage further investigation.