Dekompozicija HDsEMG signala temeljem slijepe separacije izvora

master's thesis

• Title (english): Decomposition of HDsEMG signals based on blind source separation
• Author: Šimun Krmek
• Mentor: Željka Lučev Vasić (mentor)
• Committee member: Željka Lučev Vasić (predsjednik povjerenstva)
• Committee member: Mario Cifrek (član povjerenstva)
• Committee member: Hrvoje Džapo (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-07-05, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovome radu napravljena je dekompozicija HDsEMG signala koristeći algoritam za slijepu separaciju izvora. Konkretnije, koristio se kmCKC algoritam koji je sačinjen od kombinacije CKC algoritma te algoritma k srednjih vrijednosti. Kao rezultat dekompozicije dobiju se vektori vremenskih zapisa pobuda aktivnih motoričkih jedinica. Dekompozicija danim algoritmom testirala se kroz tri scenarija u kojem se mijenjao ulazni signal. U prvom scenariju, algoritam se testirao nad nasumično generiranim signalima, u drugome slučaju ulazni su signali bili sintetički HDsEMG signali, dok su u posljednjem slučaju korišteni stvarni, snimljeni HDsEMG signali. Algoritam je u prva dva scenarija uspio, za različite kvalitete signala, identificirati gotovo većinu aktivnih izvora, odnosno motoričkih jedinica u tome trenutku. Kod scenarija sa stvarnim signalima, dekompozcija se vršila nad signalima koji su mjerili aktivnost mišića ekstenzora podlaktice prilikom pokreta prstiju. U prosjeku se prilikom svake dekompozicije identificiralo 12.4 aktivnih motoričkih jedinica. Također, prosječna stopa pobude iznosila je 12.37 pps. Dobiveni broj estimiranih motoričkih jedinica i stopa pobude u skladu je s očekivanim vrijednostima. Pored broja estimiranih izvora, algoritam je pokazao visoku osjetljivost, odnosno preciznost dekompozicije. Zaključno, isti je pokazao vrlo dobre rezultate te se smatra pogodnim za upotrebu prilikom dekompozicije HDsEMG signala.

Abstract (english)

In this paper, the HDsEMG signal was decomposed using an algorithm for blind source separation. More specifically, the kmCKC algorithm was used, which is a combination of the CKC algorithm and the k-means clustering. As a result of decomposition, firing times of active motor units are obtained. Decomposition with the given algorithm was tested through three scenarios in which the input signal was varied. In the first scenario, the algorithm was tested on randomly generated signals. In the second case, the input signals were synthetic HDsEMG signals. Finally, the algorithm was tested on real, recorded HDsEMG signals. In the first two scenarios, the algorithm was able to identify almost the majority of active sources, i.e. motor units at that moment, for different signal quality. In the scenario with real signals, the decomposition was performed on the signals that measured the activity of the forearm extensor muscles during finger movements. On average, 12.4 active motor units were identified during each decomposition. Also, the average excitation rate was 12.37 pps. The obtanied number of estimated motor units and the excitation rate is within the expected values. In addition to the number of estimated sources, the algorithm showed high sensitivity, that is, the precision of the decomposition. In conclusion, the algorithm showed very good results and is considered suitable for use in HDsEMG signal decomposition.

• Keywords: slijepa separacija izvora; kompenzacija konvolucijske jezgre uz pomoć k-srednjih vrijednosti; površinski EMG signali visoke prostorne razlučivosti; vremenski zapisi pobuda aktivnih motoričkih jedinica
• Keywords (english): blind source separation; convolutional kernel compensation with k-means clustering; high density surface EMG signals; motor unit firing times
• Language: croatian
• URN:NBN: urn:nbn:hr:168:712165
• Study programme: Title: Information and Communication Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije (magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-10-11 09:26:58