Abstract
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.
Highlights
Electrical and Computer Engineering Department, University of Tehran, Tehran 14395-515, Iran; Pacific Parkinson’s Research Centre, Djavad Mowafaghian Centre for Brain Health, University of British
We demonstrate that a latent space found by a deep transfer learning method to discriminate between HC and Parkinson’s disease (PD) proves superior to standard methods such as sparse discriminant analyses in associating the EEG and clinical features such as UPDRS3 [18]
We extract latent features of this model and demonstrate that these are more closely aligned with three important clinical indices of tremor, finger tap performance, and body bradykinesia compared to the original directional connectivity (DC) values, with each clinical feature having a unique DC
Summary
Electrical and Computer Engineering Department, University of Tehran, Tehran 14395-515, Iran; Pacific Parkinson’s Research Centre, Djavad Mowafaghian Centre for Brain Health, University of British. There have been advances in the high-accuracy performance of deep neural networks (DNNs) using few-patient data. We propose a method to transform resting-state EEG data into a deep latent space to classify PD subjects from healthy cases. PD is characterized primarily by the degeneration of dopaminergic cells in the substantia nigra in the midbrain [2], leading to dopamine depletion in the striatum [3]. This neurochemical alteration impairs neuronal processing in the basal ganglia [4], resulting in prominent 8–30 Hz oscillations in local field potential recordings
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call