Klasifikasi Covid-19 Pada Citra CT-Scan Dengan Mengombinasikan Metode Convolutional Neural Network Dan Support Vector Machine

Classification Of Covid-19 in CT-Scan Images By Combining The Convolutional Neural Network And Support Vector Machine

Covid-19 merupakan jenis penyakit menular dengan cepat yang disebabkan oleh virus corona yang mewabah pada akhir Desember 2019. Dalam mendeteksi Covid-19 perlu adanya pemeriksaan dengan tes swab, RT-PCR,  Chest X-Ray dan CT Scan. Hasil citra CT-Scan memilki kualitas jelas dibanding pemeriksaan lainnya. Untuk hasil yang lebih jelas dalam mendeteksi Covid-19 pada pasien dengan menggunakan pendekatan machine learning pada citra CT-Scan paru-paru. Metode machine learning yang digunakan yaitu Convolutional Neural Network untuk melakukan proses ekstraksi fitur dan Support Vector Machine untuk klasifikasi Covid-19. Pada tahap klasifikasi Covid-19 pada citra CT-Scan paru-paru terdapat beberapa tahapan proses utama yaitu pendeskripsian data, pra-pemrosesan, ekstraksi fitur dan klasifikasi. Kelas data yang digunakan yaitu kelas covid dan no covid.

Sehingga hasil dari proses pengujian dalam mengombinasikan metode CNN dan SVM didapatkan akurasi sebesar 97.39%, precision sebesar 97.36%, recall sebesar 97.43%, f1-score sebesar 97.39% dan auc sebesar 97.43% dengan parameter CNN yaitu optimizer adam dan parameter SVM yaitu kernel Poly, nilai Degree sebesar 5, nilai Gamma sebesar 1 dan C sebesar 10.

Kata Kunci— Machine Learning, Deep Learning, Convolutional Neural Network, Support Vector Machine, Covid-19.

Covid-19 is a type of rapidly contagious disease caused by a corona virus that was endemic at the end of December 2019. In detecting Covid-19, an examination is needed with a swab test, RT-PCR, Chest X-Ray and CT Scan. CT-Scan image results have clear quality compared to other examinations. For clearer results in detecting Covid-19 in patients using a machine learning approach on CT-scan images of the lungs. The machine learning method used is the Convolutional Neural Network to carry out the feature extraction process and the Support Vector Machine for the classification of Covid-19. In the Covid-19 classification stage on CT-scan images of the lungs, there are several main processing stages, namely data description, pre-processing, feature extraction and classification. The data classes used are the covid and no covid classes. So that the results of the testing process in combining the CNN and SVM methods obtained an accuracy of 97.39%, precision of 97.36%, recall of 97.43%, f1-score of 97.39% and auc of 97.43% with CNN parameters namely optimizer adam and SVM parameters namely kernel Poly , the Degree value is 5, the Gamma value is 1 and C is 10.

Key words : Machine Learning, Convolutional Neural Network, Support Vector Machine, Covid-19.