D the UVA Department of Otolaryngology ?Head and Neck Surgery.NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author ManuscriptAbbreviationsp70S6K PI3K RPPA mTOR MAPK EGFR NSCLC RTK FOX T-PER HNSCC PKA 5’TOP 4E-BP1 eIF4E p70S6 kinase phosphoinositide-3-kinase Reverse Phase Protein Array mammalian target of rapamycin mitogen activated protein kinase epidermal growth aspect receptor non-small cell lung cancer receptor tyrosine kinase forkhead box tissue protein extraction reagent head and neck squamous cell carcinoma protein kinase A 5′ terminal oligopyrimidine tract eukaryotic initiation aspect 4E binding protein eukaryotic initiation factor 4E
Hamedi et al. BioMedical Engineering On-line 2013, 12:73 http://biomedical-engineering-online/content/12/1/RESEARCHOpen AccessEMG-based facial gesture recognition through versatile elliptic basis function neural networkMahyar Hamedi1*, Sh-Hussain Salleh2, Mehdi Astaraki3 and Alias Mohd Noor* Correspondence: [email protected] 1 Faculty of Bioscience and Medical Engineering, Universiti Teknologi Malaysia, Skudai, Johor 81310, Malaysia Complete list of author facts is obtainable at the finish on the articleAbstractBackground: Recently, the recognition of distinct facial gestures working with facial neuromuscular activities has been proposed for human machine interfacing applications.Price of 250674-51-2 Facial electromyograms (EMGs) evaluation is a complex field in biomedical signal processing exactly where accuracy and low computational expense are considerable issues. In this paper, a very rapid versatile elliptic basis function neural network (VEBFNN) was proposed to classify distinct facial gestures.4-Bromo-6-methylpyridin-2-amine Data Sheet The effectiveness of different facial EMG time-domain features was also explored to introduce the most discriminating.PMID:34337881 Techniques: In this study, EMGs of ten facial gestures had been recorded from ten subjects working with three pairs of surface electrodes within a bi-polar configuration. The signals have been filtered and segmented into distinct portions before function extraction. Ten distinct time-domain characteristics, namely, Integrated EMG, Imply Absolute Worth, Mean Absolute Value Slope, Maximum Peak Value, Root Mean Square, Uncomplicated Square Integral, Variance, Mean Value, Wave Length, and Sign Slope Adjustments have been extracted from the EMGs. The statistical relationships involving these characteristics have been investigated by Mutual Info measure. Then, the function combinations including two to ten single capabilities had been formed primarily based on the feature rankings appointed by MinimumRedundancy-Maximum-Relevance (MRMR) and Recognition Accuracy (RA) criteria. Within the final step, VEBFNN was employed to classify the facial gestures. The effectiveness of single characteristics too because the function sets on the program functionality was examined by thinking about the two big metrics, recognition accuracy and training time. Lastly, the proposed classifier was assessed and compared with standard procedures assistance vector machines and multilayer perceptron neural network. Outcomes: The average classification final results showed that the most beneficial overall performance for recognizing facial gestures among all single/multi-features was achieved by Maximum Peak Worth with 87.1 accuracy. Moreover, the outcomes proved a very fast process since the instruction time for the duration of classification via VEBFNN was 0.105 seconds. It was also indicated that MRMR was not a right criterion to become made use of for creating extra efficient function sets in comparison with RA. Conclusions: This function was accomplished by introducing probably the most d.
