doktorska disertacija
Filip Urh (Avtor), Aleš Holobar (Mentor)

Povzetek

V doktorski disertaciji razvijemo in predstavimo algoritem WBS (ang. Witness Based Segmentation) za ocenjevanje pravilnosti razpoznave posameznih impulzov iz večkanalnih konvolutivnih mešanic impulznih izvorov. Omejimo se na signale večkanalnih površinskih elektromiogramov (ang. High-Density surface Electromyogram % HDsEMG), katerih vlaki impulzov posamezne motorične enote (ME) so ocenjeni s pomočjo dekompozicijske metode kompenzacije konvolutivnih jeder (ang. Convolution Kernel Compensation % CKC).%Ocenjevanje uspešnosti dekompozicije signalov HDsEMG je velikega pomena, ko želimo dobljene vlake impulzov uporabiti v nadaljnjih aplikacijah na področju nevrorehabilitacije, protetike, ortopedije, ergonomije, športnih znanosti, ambientne inteligence in v industriji računalniških iger. Razvitih je bilo že več metod, vsaka s svojimi prednostmi in slabostmi, vendar je vsem skupno to, da ne podpirajo ocenjevanja pravilnosti razpoznave posameznega proženja v vlaku impulzov ME, temveč podajajo skupno oceno prepoznave vseh proženj posamezne ME.%Osnovo razvitega algoritma WBS predstavlja že uveljavljena metrika razmerja impulz-šum (ang. Pulse-to-Noise Ratio % PNR), ki jo v prvem koraku razvitega algoritma nadgradimo z uporabo prič, ki jih izberemo med impulzi ME, razpoznanimi s pomočjo metode CKC. Ugotovili smo, da se vrednost metrike PNR, izračunane nad pričami, statistično značilno zniža, če v izračun filtra ME, s katerim nato ocenimo vlak impulzov, poleg prič vključimo tudi nepravilno razpoznani impulz. To lastnost uporabimo za klasifikacijo nadgrajene metrike PNR, ki nam omogoča ocenjevanje ali posamezen impulz, ki ni priča, pripada množici pravilno ali napačno razpoznanih impulzov. Strategija izbire prič, število prič in število dodanih impulzov, ki s strani dekompozicijske metode CKC niso bili segmentirani kot proženja ME, predstavljajo notranje parametre algoritma zato v doktorski disertaciji temeljito analiziramo njihov vpliv.%Uspešnost razvitega algoritma WBS preverimo na vlakih impulzov ME, pridobljenih iz sintetičnih in eksperimentalnih signalov HDsEMG s pomočjo dekompozicijske metode CKC. Sintetičnim signalom dvoglave nadlaktne mišice je bil dodan Gaussov oziroma Laplaceov šum z razmerjem signal-šum 30, 20 in 10 dB. Množica eksperimentalnih signalov je bila sestavljena iz signalov dvoglave nadlaktne mišice in prednje golenske mišice, pri katerih smo imeli poleg signalov HDsEMG tudi znotrajmišične signale EMG, ter iz signalov sočasno aktivnih mečnih mišic.%Razvit algoritem WBS ima zgornjo asimptotično mejo časovne zahtevnosti in porabe pomnilnika O(n), kar pomeni, da časovna in prostorska zahtevnost algoritma rasteta linearno s številom testiranih impulzov. Kot tak je algoritem WBS časovno in prostorsko učinkovit in tako primeren za uporabo v realnem času med urejanjem vlaka impulzov posamezne ME.%V skoraj vseh primerih signalov HDsEMG je bilo po uporabi algoritma WBS v vlaku impulzov statistično značilno manj nepravilno razpoznanih proženj kot pred njegovo uporabo. Rezultati se statistično značilno niso izboljšali le v primerih eksperimentalnih signalov brez ali z zelo malo šuma, kjer je zelo dobre rezultate vračala že dekompozicijska metoda CKC. Uspešno delovanje algoritma smo posebej preverili in potrdili tudi na vlakih impulzov ME, ki so s strani metode CKC razpoznani s slabšo kakovostjo in imajo PNR < 28 dB.

Ključne besede

večkanalni površinski elektromiogram;dekompozicija;motorična enota;vlaki impulzov;ocenjevanje pravilnosti razpoznave;posamezni impulz;konvolutivne mešanice impulznih izvorov;pravilnost dekompozicije;doktorske disertacije;

Podatki

Jezik: Slovenski jezik
Leto izida:
Tipologija: 2.08 - Doktorska disertacija
Organizacija: UM FERI - Fakulteta za elektrotehniko, računalništvo in informatiko
Založnik: [F. Urh]
UDK: 004.93.021:621.37(043.3)
COBISS: 86077699 Povezava se bo odprla v novem oknu
Št. ogledov: 508
Št. prenosov: 75
Ocena: 0 (0 glasov)
Metapodatki: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Ostali podatki

Sekundarni jezik: Angleški jezik
Sekundarni naslov: Accuracy assessment of individual pulse identification from multichannel convolutive mixtures of pulse sources
Sekundarni povzetek: In this doctoral thesis, we develop and present the Witness Based Segmentation (WBS) algorithm for accuracy assessment of individual pulse identification from multichannel convolutive mixtures of pulse sources. We limit ourselves to the High-Density surface Electromyograms (HDsEMG), whose individual motor unit (MU) pulse trains are estimated by using previously established Convolution Kernel Compensation (CKC) decomposition method.%Evaluating the performance of HDsEMG decomposition is of great importance when we use the decomposition results to support the solutions in the fields of neurorehabilitation, prosthetics, orthopedics, ergonomics, ambient intelligence, sports sciences and in the video game industry. Several accuracy assessment techniques have already been developed, each with its advantages and disadvantages. But none of them supports the accuracy assessment for every individual MU firing identification. Instead, they mostly provide a joint accuracy assessment for identification of all the MU firings of each MU.%The developed WBS algorithm builds on the already established Pulse-to-Noise Ratio (PNR) metric, which is upgraded in the first step of the developed algorithm. For this purpose, we use witnesses, selected from MU pulses that are identified by the CKC method. We demonstrate that the value of the PNR metric calculated on these witnesses decreases statistically significantly when an incorrectly identified pulse is included in the calculation of the MU filter used to estimate the MU pulse train. We then use this upgraded PNR metric to automatically estimate whether an individual pulse that is not a witness belongs to the set of correctly or incorrectly identified MU firings. The witness selection strategy, the number of witnesses, and the number of added pulses, which were not identified as MU firings by the CKC decomposition method, represent the internal parameters of the algorithm. Therefore, we thoroughly analyze their impact on the performance of the WBS algorithm.%The performance of the developed WBS algorithm is tested on MU pulse trains identified by the CKC decomposition method from both synthetic and experimental HDsEMG signals. Gaussian and Laplacian noise with a signal-to-noise ratio of 30, 20 and 10 dB was added to the synthetic signals of the simulated biceps brachii muscle. The set of experimental signals consisted of signals from biceps brachii and tibialis anterior muscles and signals from simultaneously active triceps surae muscles. In the case of biceps brachii and tibialis anterior muscle, we also had intramuscular EMG signals acquired simultaneously with the HDsEMG signals. Therefore, we used the previously established two-source validation method to identify the currently and incorrectly identified MU firings from HDsEMG signals and validate our WBS algorithm. %The developed WBS algorithm exhibits linear computation complexity O(n), which means that its complexity increases linearly with the number of tested MU pulses. As such, the WBS algorithm is time and memory efficient and suitable also for use in real time MU firings identification.%The efficiency of the WBS algorithm is well demonstrated by the results presented in this thesis. In almost all the tested cases, there were significantly fewer incorrectly identified MU firings after the application of the algorithm than before its use. The results did not improve only in the case of experimental signals without or with very little noise, where the CKC decomposition method already returned high quality results. The efficiency of the algorithm was most evident in the cases of MUs that were identified by the CKC method with poor quality, that is in MUs that were identified by CKC method with PNR value lower than 28 dB.
Sekundarne ključne besede: high-density surface electromyogram;decomposition;motor unit;pulse trains;identification accuracy assessment;individual firing;convolutive mixtures of pulse sources;decomposition accuracy;Matematična morfologija;
Vrsta dela (COBISS): Doktorsko delo/naloga
Komentar na gradivo: Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko
Strani: VIII, 214 str.
ID: 12865043