Abstract
This paper proposes two new data augmentation approaches based on Deep Convolutional Generative Adversarial Networks (DCGANs) and Style Transfer for augmenting Parkinson’s Disease (PD) electromyography (EMG) signals. The experimental results indicate that the proposed models can adapt to different frequencies and amplitudes of tremor, simulating each patient’s tremor patterns and extending them to different sets of movement protocols. Therefore, one could use these models for extending the existing patient dataset and generating tremor simulations for validating treatment approaches on different movement scenarios.
Highlights
During the evaluation of the model, we have identified that the competition between the content loss function (Lcont ) and the style loss function (Lsty ) made it impossible for the optimizer to create an output with higher amplitudes of tremor on the static part of the signals, since Lcont tries to keep the amplitudes close to the content signal by the nature of the mean squared error (MSE)
This paper proposes two new data augmentation methods for EMG signal generation using
Deep Convolutional Generative Adversarial Networks (DCGANs) and Style Transfer, creating a reference implementation based on Python
Summary
As one of the most common neurodegenerative diseases that affects approximately 10 million people around the world [1], Parkinson’s Disease (PD) has been studied and investigated from different manners and perspectives, in order to minimize the disease’s symptoms and impairments to patients.Many studies around rest and action tremors have been conducted, whereas surface electromyography (sEMG) stands out as one of the most common ways to measure muscle response to voluntary or involuntary stimulation, being widely used as main input and feedback signal for artificial stimulation devices [2,3,4].EMG is widely used clinically for the diagnosis of neurological and muscular pathology [5], and has recently been used for several human–machine interface applications, such as controlling computer interfaces, navigation through virtual reality environments, controlling robots, drones, and other interesting applications [6].acquiring such datasets from patients is a complicated and sometimes painful task.Most patients that experience unpleasant effects during such experiments, such as tiredness, fatigue [7], and a wide range of movements, are usually not possible due to the patient’s movement limitation and impairment due to the disease.collecting, processing, and using recorded EMG signals for analysis is quite a challenging approach, due to data scarcity and lack of dataset variation. EMG is widely used clinically for the diagnosis of neurological and muscular pathology [5], and has recently been used for several human–machine interface applications, such as controlling computer interfaces, navigation through virtual reality environments, controlling robots, drones, and other interesting applications [6]. Acquiring such datasets from patients is a complicated and sometimes painful task. Collecting, processing, and using recorded EMG signals for analysis is quite a challenging approach, due to data scarcity and lack of dataset variation. Data augmentation is a promising alternative approach for extending existing datasets, which could allow further research and analysis
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
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 callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
