Samudra, Bayu (2018) Support Vector Machine untuk Deteksi Kejang Epilepsi Berdasarkan Sinyal EEG. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
06211440000115-Undergraduate_Theses.pdf - Accepted Version Download (2MB) | Preview |
Abstract
Gangguan saraf yang paling umum ditemui salah satunya adalah epilepsi. Epilepsi merupakan penyakit yang dapat menyerang siapapun, tanpa melihat jenis kelamin ataupun usia. Diagnosis yang tepat dan penanganan yang tepat waktu terhadap epilepsi merupakan hal yang penting bagi pasien. Ananlisis EEG adalah metode yang umum dilakukan untuk mendiagnosa epilepsi, tetapi inspeksi visual oleh ahli saraf membutuhkan waktu yang relatif lama . Penelitian ini akan dilakukan deteksi kejang epilepsi dengan metode machine learning berdasarkan sinyal EEG. Data EEG yang digunakan pada tugas akhir ini berasal dari database yang tersedia di website milik Klinik für Epileptologie, Universität Bonn. Metode Discrete wavelet transform digunakan untuk mendekomposisi sinyal menjadi pita-pita frekuensi. Pita-pita frekuensi tersebut selanjutnya diekstrasi fitur energy, variance, entropy dan line length. Setelah didapatkan fitur-fitur pada setiap pita frekuensi selanjutnya dilakukan klasifikasi dengan metode Support vector machine (SVM) menggunakan kernel linear dan RBF. Hasil analisa menunjukkan bahwa metode SVM menggunakan kernel linear dan rbf memiliki hasil akurasi, spesitivitas, sensitivitas dan AUC masing-masing sebesar 98,4%, 98,75%, 97% dan 97,875%.
=============================================================The most common neurological disorders encountered worldwide are epilepsy. Epilepsy is a disease that can attack anyone, regardless of gender or age. Two of the three patients were able to control the seizures satisfactorily with anti-seizure medication. EEG analysis is a common method for diagnosing epilepsy. The data used in this thesis comes from the database avaliable on Klinik für Epileptologie, Universität Bonn website. The data amounts to 500 consisted of five sets of epilepsy seizures (set S) and not epileptic seizures (sets Z, N, O, F). Discrete wavelet transform method is used to decompose the signal into frequency bands. The bands are then extracted into features of energy, variance, entropy and line length. After obtaining the features of each frequency band, the following will be classified using the Support vector machine (SVM) method using linear kernel and rbf. The result of analysis shows that SVM method using linear kernel and RBF has accuracy, specificity, sensitivity and AUC of 98.4%, 98.75%, 97% and 97,875% respectively.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Uncontrolled Keywords: | Accuracy, AUC, Discrete Wavelet Transform, Electroencephalographic, Epilepsi, Sensitivity, Specitifity |
| Subjects: | Q Science > QA Mathematics > QA403.3 Wavelets (Mathematics) R Medicine > RC Internal medicine > RC0321 Neuroscience. Biological psychiatry. Neuropsychiatry |
| Divisions: | Faculty of Mathematics and Science > Statistics > 49201-(S1) Undergraduate Thesis |
| Depositing User: | Bayu samudra |
| Date Deposited: | 08 Jul 2021 09:17 |
| Last Modified: | 08 Jul 2021 09:17 |
| URI: | http://repository.its.ac.id/id/eprint/57026 |
Actions (login required)
 |
View Item |