Active Research Works

Developing Deep Learning Algorithms to Diagnose Subtypes of Depression
This project, in collaboration with the Department of Psychiatry and Behavioral Neuroscience, McMaster University, explores a novel deep learning algorithm (DLA) based on effective connectivity between brain’s regions that are extracted from resting electroencephalography (EEG) data to diagnose different subtypes of depressions including major depressive disorder, depressive episode of bipolar disorder, manic episode of bipolar disorder, atypical disorder, and psychotic disorder as well as schizophrenia.

Developing EEG Biomarkers to Predict Response to Antidepressant Medications in Major Depressive Disorder (MDD)

In this project, in collaboration with the Department of Psychiatry and Behavioral Neurosciences, McMaster University, researchers are developing a deep learning algorithm (DLA) based on our newly developed robust brain source localization technique and the effective connectivity to predict response to three medications Sertraline, Bupropion, and Placebo based on pre-treatment electroencephalography (EEG) data. In addition, the same methodology is used to predict suicide ideation.  

Developing Quantitative Sensing Technologies to Measure the Fine Motor Skills and Evaluate the Efficacy of Therapeutic Interventions for Autistic Children 

In this study, in collaboration with the Department of Occupational Therapy, New York Tech, researchers are developing sensing technologies that can quantitatively measure the patterns of the fine motor activities that can be used to 1) improve fine motor skills and 2) evaluate the efficacy of the outcomes of the therapeutic intervention. The inexpensive, non-invasive, and accessible tools, designed through this technology, can assist ASD children to improve their motor skills and help the clinician to better evaluate the efficacy of the intervention.

EEG Biomarkers to Investigate the Effect of Vagus Nerve Stimulation (VNS) on the Brain’s Functional Activity using Vagal-Evoked Potential

Vagus nerve stimulation (VNS) is a nonpharmacological treatment for epilepsy and depression and has been tested as a possible therapy for tinnitus, post-traumatic stress disorder, headache, sleep disorders, and neurorehabilitation after stroke. However, the cortical effects of VNS, manifesting as vagal-evoked potentials (VEPs) at the time of stimulation on modulating the brain’s activity is unknown. Therefore, in this project, in collaboration with the Institute of Bioelectronic Medicine, Feinstein Institutes for Medical Research, researchers plan to find quantitative biomarkers based on advanced signal processing approaches to investigate modulations in brain’s electrical activity as a result of VEPs. This has implications for the role of cortical responses to stimuli in ongoing cortical activity, and for the dose calibration of VNS therapies.

Near-Field Holographic Microwave Imaging Augmented with Convolutional Neural Network

In this project we plan to develop a method to incorporate a three-dimensional convolution neural network (CNN) to obtain prior knowledge about the azimuthal position of the imaged object in a circular near-field holographic microwave imaging (NH-MWI) system when limited data is acquired in terms of the number of frequencies and the number of receivers. By using the scattered responses of various objects within a circular scanned region and considering their azimuthal locations as targets, a training set is created for the utilized CNN. The predicted azimuthal location by CNN is then used to solve the NH-MWI in the frequency domain through regularization and optimization. This approach significantly reduces the number of antennas required for data collection while illuminating the object using a single-frequency signal to provide range (radial) resolution.