Parkinson’s disease (PD) is characterized by abnormal brain oscillations that can change rapidly. Tracking neural alternations with high temporal resolution electrophysiological monitoring methods such as EEG can lead to valuable information about alterations observed in PD. Concomitantly, there have been advances in the high-accuracy performance of deep neural networks (DNNs) using few-patient data. In this study, we propose a method to transform resting-state EEG data into a deep latent space to classify PD subjects from healthy cases. We first used a general orthogonalized directed coherence (gOPDC) method to compute directional connectivity (DC) between all pairwise EEG channels in four frequency bands (Theta, Alpha, Beta, and Gamma) and then converted the DC maps into 2D images. We then used the VGG-16 architecture (trained on the ImageNet dataset) as our pre-trained model, enlisted weights of convolutional layers as initial weights, and fine-tuned all layer weights with our data. After training, the classification achieved 99.62% accuracy, 100% precision, 99.17% recall, 0.9958 F1 score, and 0.9958 AUC averaged for 10 random repetitions of training/evaluating on the proposed deep transfer learning (DTL) network. Using the latent features learned by the network and employing LASSO regression, we found that latent features (as opposed to the raw DC values) were significantly correlated with five clinical indices routinely measured: left and right finger tapping, left and right tremor, and body bradykinesia. Our results demonstrate the power of transfer learning and latent space derivation for the development of oscillatory biomarkers in PD.

Interaction of electromagnetic (EM) waves with human tissue has been a longstanding research topic for electrical and biomedical engineers. However, few numbers of publications discuss the impacts of external EM-waves on neural stimulation and communication through the nervous system. In fact, complex biological neural channels are a main barrier for intact and comprehensive analyses in this area. One of the everpresent challenges in neural communication responses is dependency of vesicle release probability on the input spiking pattern. In this regard, this study sheds light on consequences of changing the frequency of external EM-wave excitation on the post-synaptic neuron's spiking rate. It is assumed that the penetration depth of the wave in brain does not cover the postsynaptic neuron. Consequently, we model neurotransmission of a bipartite chemical synapse. In addition, the way that external stimulation affects neurotransmission is examined. Unlike multiple frequency component EM-waves, the monochromatic incident wave does not face frequency shift and distortion in dispersive media. In this manner, a single frequency signal is added as external current in the modified leaky integrated-andfire (LIF) model. The results demonstrate existence of a node equilibrium point in the first order dynamical system of LIF model. A fold bifurcation (for presupposed LIF model values) occurs when the external excitation frequency is near 200 Hz. The outcomes provided in this paper enable us to select proper frequency excitation for neural signaling. Correspondingly, the cut-off frequency reliance on elements' values in LIF circuit is found.