Deteksi Tingkat Keparahan Retinopati Diabetik Menggunakan EfficientNet V2-S
Keywords:
Diabetic Retinopathy, EfficientNetV2-S, Deep Learning, Transfer Learning, Computer VisualAbstract
Retinopati diabetik merupakan komplikasi mikrovaskular diabetes melitus yang menjadi penyebab utama kebutaan pada populasi usia produktif, sehingga deteksi dini melalui screening rutin sangat krusial untuk mencegah kehilangan penglihatan permanen. Penelitian ini mengembangkan sistem deteksi otomatis retinopati diabetik menggunakan arsitektur deep learning EfficientNetV2-S dengan pendekatan klasifikasi biner untuk membedakan citra retina normal dan yang terindikasi retinopati diabetik. Dataset APTOS 2019 yang terdiri dari 3662 citra fundus retina dipreprocessing dan dibagi menjadi training set sebanyak 2929 citra dan validation set sebanyak 733 citra dengan stratified splitting rasio 80:20 untuk memastikan distribusi kelas yang proporsional. Model diimplementasikan menggunakan framework PyTorch dengan strategi transfer learning dari bobot pre-trained ImageNet. Training dilakukan selama 20 epoch dengan batch size 16 menghasilkan performa optimal pada epoch ke-8 dengan validation accuracy 99.32 persen, precision 99.32 persen, recall 99.32 persen, dan F1-score 99.32 persen. Confusion matrix model memiliki sensitivitas yang sangat tinggi dengan hanya 3 false negative dari 372 kasus true positive dan spesifisitas bagus dengan hanya 2 false positive dari 361 kasus false negative. Hasil penelitian mendemonstrasikan bahwa EfficientNetV2-S sangat bagus sebagai solusi screening automated yang cost-effective untuk early detection retinopati diabetik, terutama untuk implementasi pada telemedicine platforms di daerah dengan keterbatasan akses specialist oftalmologist.
References
X. Wang, C. Qiu, X. Ren, Z. Xiong, V. C. M. Leung, and D. Niyato, "Research Challenges and Future Directions," Wirel. Networks (United Kingdom), pp. 105–108, 2022, doi: 10.1007/978-3-031-10186-1_6.
R. Amelia, M. D. Sari, V. Virgayanti, H. Wijaya, R. M. Situmorang, and S. Nadyah, "Clinical assessment of retinopathy in type 2 diabetes: a cross-sectional study," Fam. Med. Prim. Care Rev., vol. 26, no. 4, pp. 433–437, 2024, doi: 10.5114/fmpcr.2024.144910.
dr. S. Evani, "Deteksi Retinopati Diabetik dengan Artificial Intelligence," https://www.alomedika.com. Accessed: Nov. 02, 2025. [Online]. Available: https://www.alomedika.com/deteksi-retinopati-diabetik-dengan-ai
Maudisha, "Preventing Diabetic Retinopathy in Diabetic Patients, FKUI Educated Health Workers and Communities on Kepulauan Seribu," www.ui.ac.id. [Online]. Available: https://www.ui.ac.id/en/preventing-diabetic-retinopathy-in-diabetic-patients/
É. B. Rangel, C. O. Rodrigues, and J. R. De Sá, "Micro- and Macrovascular Complications in Diabetes Mellitus: Preclinical and Clinical Studies," J. Diabetes Res., vol. 2019, no. 1, pp. 96–123, 2019, doi: 10.1155/2019/2161085.
C. Suedumrong, S. Phongmoo, T. Akarajaka, and K. Leksakul, "Diabetic Retinopathy Detection Using Convolutional Neural Networks with Background Removal, and Data Augmentation," Appl. Sci., vol. 14, no. 19, 2024, doi: 10.3390/app14198823.
K. F. Manampiring and U. M. Nusantara, "Gambar 2.1. Arsitektur CNN Sumber: [10]," pp. 7–24.
D. S. Candra, "Implementasi Convolutional Neural Network (CNN) Untuk Klasifikasi Citra Bunga," vol. 16, no. 1, pp. 2580–2582, 2020.
U. Bhimavarapu and G. Battineni, "Deep Learning for the Detection and Classification of Diabetic Retinopathy with an Improved Activation Function," Healthc., vol. 11, no. 1, 2023, doi: 10.3390/healthcare11010097.
S. Akhtar et al., "A deep learning based model for diabetic retinopathy grading," Sci. Rep., vol. 15, no. 1, pp. 1–20, 2025, doi: 10.1038/s41598-025-87171-9.
H. Liu, K. Yue, S. Cheng, C. Pan, J. Sun, and W. Li, "Hybrid model structure for diabetic retinopathy classification," J. Healthc. Eng., vol. 2020, 2020, doi: 10.1155/2020/8840174.
M. Azmoodeh-Kalati, H. Shabani, M. S. Maghareh, Z. Barzegar, and R. Lashgari, "Leveraging an ensemble of EfficientNetV1 and EfficientNetV2 models for classification and interpretation of breast cancer histopathology images," Sci. Rep., vol. 15, no. 1, pp. 1–25, 2025, doi: 10.1038/s41598-025-06853-6.
S. Tummala, V. S. G. Thadikemalla, S. Kadry, M. Sharaf, and H. T. Rauf, "EfficientNetV2 Based Ensemble Model for Quality Estimation of Diabetic Retinopathy Images from DeepDRiD," Diagnostics, vol. 13, no. 4, 2023, doi: 10.3390/diagnostics13040622.
Y. NA, "Efficientnet v2-s Architecture," https://www.researchgate.net. Accessed: Nov. 02, 2025. [Online]. Available: https://www.researchgate.net/figure/Efficientnet-v2-s-Architecture-a-represents-the-schematic-of-the-model-from-image-input_fig4_373782480
S. Kevin, S. K. Singh, M. P. Kadam, and M. S. Patil, "RESEARCH IN ENGINEERING MANAGEMENT CATARACT EYE DETECTION USING DEEP LEARNING MODELS," pp. 1092–1101, 2025.
J. G. Purnomo, S. Birowo, and M. A. Maulana, "Identifikasi Malaria Pada Citra Darah Dengan Convolutional Neural Network," bit-Tech, vol. 7, no. 2, pp. 406–412, 2024, doi: 10.32877/bt.v7i2.1828.
M. Tan and Q. V. Le, "EfficientNetV2: Smaller Models and Faster Training," Proc. Mach. Learn. Res., vol. 139, pp. 10096–10106, 2021.
S. Karthik, M., & Dane, "APTOS 2019 Blindness Detection," https://www.kaggle.com. Accessed: Feb. 07, 2025. [Online]. Available: https://kaggle.com/competitions/aptos2019-blindness-detection
A. Noriega et al., "Screening diabetic retinopathy using an automated retinal image analysis system in independent and assistive use cases in mexico: Randomized controlled trial," JMIR Form. Res., vol. 5, no. 8, pp. 1–12, 2021, doi: 10.2196/25290.
