Tristianti, Novita (2017) Klasifikasi Gerakan Otot Lengan Bawah pada Penderita Stroke Berdasarkan Sinyal EMG Menggunakan Metode K-Nearest Neighbor. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
2913100010-Undergraduate_Thesis.pdf - Published Version Download (16MB) | Preview |
Abstract
Sinyal Electromyograph (EMG) merupakan sinyal listrik yang dihasilkan oleh otot saat berkontraksi maupun berelaksasi. Saat ini, sinyal EMG banyak dikembangkan sebagai media kontrol perangkat prosthetics dan electric device. Penderita stroke memiliki keterbatasan gerak, dapat memanfaatkan sinyal EMG sebagai media kontrol untuk membantu aktivitas sehari-hari. Tugas akhir ini bertujuan mengklasifikasikan respon otot lengan bawah menggunakan metode klasifikasi K-Nearest Neighbor untuk dapat diimplementasikan pada penderita stroke sebagai media kontrol. Data sinyal EMG yang diklasifikasikan diperoleh dari dua responden normal untuk melakukan gerakan wave left dan wave right dengan enam fitur time domain MAV, VAR, RMS, IEMG, WL dan WAMP. Akurasi klasifikasi didapat rata-rata sebesar 100% pada setiap sample dengan nilai k bilangan ganjil 3 sampai 11, kemudian dilakukan pengujian untuk mengetahui performanya. Pengujian dilakukan dengan mengklasifikasi sinyal EMG dua sample penderita stroke, menambah variasi gerakan dan nilai k. Hasilnya kondisi responden dan variasi gerakan mempengaruhi nilai akurasi klasifikasi. Klasifikasi dua gerakan pada kedua responden sebesar 100%. Saat lima gerakan diklasifikasikan dengan nilai k bilangan ganjil 3-15,akurasi menurun pada kedua sample penderita stroke, sample pertama 86% dan sample
kedua 82% sedangkan pada orang normal penurunan akurasi pada sample pertama menjadi 96% dan 91% pada sample kedua. Yang terakhir adalah dengan mengklasifikasikan sepuluh gerakan, pada orang normal sampel pertama 85%, sampel kedua 81% dan pada penderita stroke sample pertama 84% sampel kedua 66%. Semakin besar nilai k, akurasi mengalami penurunan performansi.
=================================================================
Electromyograph signal (EMG) is an electrical signal generated by the muscles when contracting or relaxing. Currently, EMG signal is widely developed as a media control prosthetics device and electric devices. Stroke patients who have limited muscular nervous system movement, can using of EMG signal as a media control to facilitate daily activities. This final project purposed to classify the muscular response of the lower arms using the K-Nearest Neighbor classification method to be implemented for stroke patients as control media. EMG signal data which classified were obtained from
two sample of patient who didnt have stroke to performing left and right motions with six time domain features MAV, VAR, RMS, IEMG, WL and WAMP. Classification accuracy obtained average in 100% in each sample with the value of k odd number 3 to 11 by testet for performance. This test doing by classifying EMG signal stroke sufferers, add motion variations and value of k. The result of the respondent's condition and the variation of movement is affected the value of classification accuracy. Classification two movements of both stroke patient in 100%. When the five movements were classified by the odd number k of 3-15, the accuracy decreased in both stroke patient. The first sample decreased 86% and the second sample decreased 82%, while in normal people the accuracy decreased 96% for the first respondent and 91% for to 96% for the second respondent. The last is to classify ten movements, in normal people the first sample is 85%, the second sample is 81% and in the first stroke patient the first 84% sample is 66%. The biggest the value of k, the accuracy has decreased performance.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSKom 616.74 Tri k |
| Uncontrolled Keywords: | Electromyograph, K-Nearest Neighbor Classifier, Signal Classification, Elektromiograf, Klasifikasi sinyal |
| Subjects: | Q Science > QA Mathematics > QA76 Computer software Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) Q Science > QM Human anatomy |
| Divisions: | Faculty of Information Technology > Computer Engineering |
| Depositing User: | Novita Tristianti . |
| Date Deposited: | 26 Oct 2017 04:00 |
| Last Modified: | 05 Mar 2019 03:09 |
| URI: | http://repository.its.ac.id/id/eprint/48684 |
Actions (login required)
 |
View Item |