Exploiting the synchronized assessment of the neuromuscular implications, this paper proposes an embedded digital architecture for the assessment of the movements’ automatism and the reduction of pre-motor function capability. The study can enable a forward recognition of the Parkinson’s disease (PD) progression stages, which are characterized by muscular disorders. The architecture, implemented on Altera Cyclone V FPGA, classifies in real-time these physiological disorders during the walk. The system operates on 8 surface EMG (limbs) and 7 EEG (motor-cortex). The signals, synchronously acquired and processed, undergo to a features extraction (FE) in the time-frequency domains. The features are time-continuously processed (in chronological order) from an innovative on-going Support Vector Machine (SVM) classifier. The SVM identifies and categorizes the patient pathology severity. Experimental results from 4 subjects affected by mild (n = 2) and heavy PD (n = 2) show an accuracy 93.97% ± 2.1% in PD stages recognition.
Neuromuscular Disorders Assessment by FPGA-Based SVM Classification of Synchronized EEG/EMG / De Venuto, Daniela; Mezzina, Giovanni (LECTURE NOTES IN ELECTRICAL ENGINEERING). - In: Applications in Electronics Pervading Industry, Environment and Society : APPLEPIES 2018 / [a cura di] Sergio Saponara, Alessandro De Gloria. - ELETTRONICO. - Cham, CH : Springer, 2019. - ISBN 978-3-030-11972-0. - pp. 37-44 [10.1007/978-3-030-11973-7_5]
Neuromuscular Disorders Assessment by FPGA-Based SVM Classification of Synchronized EEG/EMG
De Venuto, Daniela
;Mezzina, Giovanni
2019-01-01
Abstract
Exploiting the synchronized assessment of the neuromuscular implications, this paper proposes an embedded digital architecture for the assessment of the movements’ automatism and the reduction of pre-motor function capability. The study can enable a forward recognition of the Parkinson’s disease (PD) progression stages, which are characterized by muscular disorders. The architecture, implemented on Altera Cyclone V FPGA, classifies in real-time these physiological disorders during the walk. The system operates on 8 surface EMG (limbs) and 7 EEG (motor-cortex). The signals, synchronously acquired and processed, undergo to a features extraction (FE) in the time-frequency domains. The features are time-continuously processed (in chronological order) from an innovative on-going Support Vector Machine (SVM) classifier. The SVM identifies and categorizes the patient pathology severity. Experimental results from 4 subjects affected by mild (n = 2) and heavy PD (n = 2) show an accuracy 93.97% ± 2.1% in PD stages recognition.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
