Perbandingan efficientnet-B0 Pretrained dan Prototypical Network from Scratch untuk Klasifikasi Diabetic Retinopathy
Keywords:
Diabetic Retinopathy, Transfer Learning, EfficientNet, Prototypical Network, Citra MedisAbstract
Penelitian ini membahas perbandingan kinerja model deep learning berbasis transfer learning dan model yang dilatih dari awal dalam klasifikasi Diabetic Retinopathy menggunakan citra fundus retina. Permasalahan utama dalam pengembangan sistem klasifikasi citra medis adalah keterbatasan data berlabel yang dapat memengaruhi kemampuan generalisasi model. Penelitian ini bertujuan menganalisis efektivitas EfficientNet-B0 pretrained dibandingkan dengan Prototypical Network from scratch pada dataset APTOS 2019 yang terdiri dari 3.662 citra fundus retina dengan lima tingkat keparahan dan distribusi kelas tidak seimbang. Metode penelitian meliputi preprocessing berupa penyesuaian ukuran citra menjadi 224×224 piksel dan normalisasi nilai piksel. EfficientNet-B0 menggunakan bobot pretrained ImageNet dengan proses fine tuning, sedangkan Prototypical Network dilatih sepenuhnya dari awal menggunakan pendekatan pembelajaran berbasis metrik. Evaluasi dilakukan menggunakan accuracy, precision, recall, F1-score, dan Quadratic Weighted Kappa. Hasil penelitian menunjukkan EfficientNet-B0 pretrained memperoleh accuracy 80,35 persen, precision 0,6587, recall 0,6279, F1-score 0,6396, dan QWK 0,8529. Sementara itu, Prototypical Network from scratch mencapai accuracy 64,53 persen, precision 0,4806, recall 0,4784, F1-score 0,4658, dan QWK 0,5712. Perbedaan performa sebesar 15,82 persen pada accuracy dan 0,2817 pada QWK menunjukkan bahwa transfer learning memberikan hasil yang lebih optimal dibandingkan pelatihan dari awal. Berdasarkan hasil tersebut, penggunaan model pretrained terbukti efektif untuk klasifikasi Diabetic Retinopathy pada kondisi keterbatasan data dan direkomendasikan dalam pengembangan sistem Computer Aided Diagnosis.
Downloads
References
Ahmed, F. (2025). Addressing High Class Imbalance in Multi-Class Diabetic Retinopathy Severity Grading with Augmentation and Transfer Learning. http://arxiv.org/abs/2507.17121
Arora, L., Singh, S. K., Kumar, S., Gupta, H., Alhalabi, W., Arya, V., Bansal, S., Chui, K. T., & Gupta, B. B. (2024). Ensemble deep learning and EfficientNet for accurate diagnosis of diabetic retinopathy. Scientific Reports, 14(1), 30554. https://doi.org/10.1038/s41598-024-81132-4
Chong, D. D., Das, N., & Singh, R. P. (2024). Diabetic retinopathy: Screening, prevention, and treatment. Cleveland Clinic Journal of Medicine, 91(8), 503–510. https://doi.org/10.3949/ccjm.91a.24028
Gu, C., & Lee, M. (2024). Deep Transfer Learning Using Real-World Image Features for Medical Image Classification, with a Case Study on Pneumonia X-ray Images. Bioengineering, 11(4), 406. https://doi.org/10.3390/bioengineering11040406
Hellín, C. J., Olmedo, A. A., Valledor, A., Gómez, J., López-Benítez, M., & Tayebi, A. (2024). Unraveling the Impact of Class Imbalance on Deep-Learning Models for Medical Image Classification. Applied Sciences, 14(8), 3419. https://doi.org/10.3390/app14083419
Kadhim, Y. A., Khan, M. U., & Mishra, A. (2022). Deep Learning-Based Computer-Aided Diagnosis (CAD): Applications for Medical Image Datasets. Sensors, 22(22), 8999. https://doi.org/10.3390/s22228999
Kropp, M., Golubnitschaja, O., Mazurakova, A., Koklesova, L., Sargheini, N., Vo, T.-T. K. S., de Clerck, E., Polivka, J., Potuznik, P., Polivka, J., Stetkarova, I., Kubatka, P., & Thumann, G. (2023). Diabetic retinopathy as the leading cause of blindness and early predictor of cascading complications—risks and mitigation. EPMA Journal, 14(1), 21–42. https://doi.org/10.1007/s13167-023-00314-8
Liu, Y., Shi, D., & Lin, H. (2024). Few-shot learning with representative global prototype. Neural Networks, 180, 106600. https://doi.org/10.1016/j.neunet.2024.106600
Ouahab, A., & Ahmed, O. Ben. (2025). ProtoMed: Prototypical networks with auxiliary regularization for few-shot medical image classification. Image and Vision Computing, 154, 105337. https://doi.org/10.1016/j.imavis.2024.105337
Pachetti, E., & Colantonio, S. (2024). A systematic review of few-shot learning in medical imaging. Artificial Intelligence in Medicine, 156, 102949. https://doi.org/10.1016/j.artmed.2024.102949
Snell, J., Swersky, K., & Zemel, R. S. (2017). Prototypical Networks for Few-shot Learning. http://arxiv.org/abs/1703.05175
Srinivasan, S., Suresh, S., Chendilnathan, C., Prakash V, J., Sivaprasad, S., Rajalakshmi, R., Anjana, R. M., Malik, R. A., Kulothungan, V., Raman, R., & Bhende, M. (2023). Inter-observer agreement in grading severity of diabetic retinopathy in wide-field fundus photographs. Eye, 37(6), 1231–1235. https://doi.org/10.1038/s41433-022-02107-1
Tan, M., & Le, Q. V. (2020). EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks. http://arxiv.org/abs/1905.11946
Teo, Z. L., Tham, Y.-C., Yu, M., Chee, M. L., Rim, T. H., Cheung, N., Bikbov, M. M., Wang, Y. X., Tang, Y., Lu, Y., Wong, I. Y., Ting, D. S. W., Tan, G. S. W., Jonas, J. B., Sabanayagam, C., Wong, T. Y., & Cheng, C.-Y. (2021). Global Prevalence of Diabetic Retinopathy and Projection of Burden through 2045. Ophthalmology, 128(11), 1580–1591. https://doi.org/10.1016/j.ophtha.2021.04.027
Widyaya, J. E., & Budi, S. (2021). Pengaruh Preprocessing Terhadap Klasifikasi Diabetic Retinopathy dengan Pendekatan Transfer Learning Convolutional Neural Network. Jurnal Teknik Informatika Dan Sistem Informasi, 7(1). https://doi.org/10.28932/jutisi.v7i1.3327
Zhou, S. K., Greenspan, H., Davatzikos, C., Duncan, J. S., Van Ginneken, B., Madabhushi, A., Prince, J. L., Rueckert, D., & Summers, R. M. (2021). A Review of Deep Learning in Medical Imaging: Imaging Traits, Technology Trends, Case Studies With Progress Highlights, and Future Promises. Proceedings of the IEEE, 109(5), 820–838. https://doi.org/10.1109/JPROC.2021.3054390
Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences. Lawrence Erlbaum Associates.
Karthik; Maggie; Dane, S. (2019). APTOS 2019 Blindness Detection. Kaggle. https://www.kaggle.com/competitions/aptos2019-blindness-detection
