Diabetic Retinopathy: Identification and Classification using Different Kernel on Support Vector Machine

Foeady, Ahmad Zoebad; Novitasari, Dian Candra Rini; Asyhar, Ahmad Hanif

doi:10.5220/0008517400720079

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Hasil deteksi citra X-ray yang terdeteksi benar pada kelasnya maupun terdeteksi salah pada kelasnya dapat dilihat pada Gambar 3. Hasil terbaik yang diperoleh pada penelitian ini sesuai dengan penelitian yang dilakukan oleh (Foeady et al, 2020) terkait klasifikasi Diabetik Retinopati menggunakan GLCM dan Kernel yang berbeda pada SVM, di mana hasil terbaik didapatkan pada kernel polynomial. Kekurangan dari penelitian yag dilakukan oleh (Foeady et al, 2020) yaitu tidak menyebutkan hasil dari sensitivitas dan spesifisitas, maka pada penelitian ini menyebutkan nilai dari akurasi, sensitivitas, dan spesifisitas.…”

Section: Hasil Dan Pembahasanunclassified

Untitled

Utami,

Meilani,

Fuad

2022

JMAT

View full text Add to dashboard Cite

Detection of Covid-19 using X-ray image data has advantages, namely, an affordable price with a high sensitivity value when compared to PCR, Swab, Antigen, and Genose tests. Therefore, this study aims to detect whether a patient is exposed to Covid-19 or not, based on lung X-ray image data using the Gray-Level Co-occurrence matrix (GLCM) feature extraction and the detection method using the Support Vector Machine (SVM). The evaluation model used in this study is the confusion matrix. This study uses 600 lung X-ray image data, consisting of 300 Covid-19 data and 300 normal data. The trials carried out in this study were using four angle orientations, namely 0°, 45°, 90°, and 135°. In addition, three types of kernels were tested, namely linear, polynomial, and RBF kernels. Data sharing uses k-fold cross validation with k=10. The best results were obtained from trials at 45° and 90° angle orientations using a polynomial kernel. The results of sensitivity, accuracy, and specificity obtained have the same value, namely 96.7% respectively. This system has an error of 3.3%. Where there is one normal Covid-19 data detected and one normal Covid-19 data detected, because the accuracy value exceeds 90.0%, it is concluded that the system built is good in detecting Covid-19 disease.

show abstract

Section: Hasil Dan Pembahasanunclassified