KLASIFIKASI PENYAKIT MATA MELALUI CITRA OPTICAL COHERENCE TOMOGRAPHY (OCT) MENGGUNAKAN METODE RESNET 101 (RESIDUAL NETWORK 101)

The eye is one of the five human senses that functions for vision, with the
process starting from the reflection of light on objects that enter through the pupil
and lens, then forwarded to the cornea and finally to the brain to be interpreted.
With the eyes, humans obtain 80% of visual information, so it is important to
maintain eye health by consuming nutritious foods, exercising, and avoiding
exposure to ultraviolet light from electronic devices. One common eye disease is
age-related macular degeneration (AMD), which can cause serious conditions
such as drusen, choroidal neovascularization (CNV), and diabetic macular edema
(DME), all of which can cause vision loss or blindness if left untreated. Detection
of eye disease can be done through Optical Coherence Tomography (OCT)
examination, which uses near-infrared light to obtain images and analyze eye
conditions accurately. In this study, the deep learning method was used, especially
ResNet 101, which is a development of the Convolutional Neural Network (CNN)
to classify OCT images of the eye. ResNet 101 involves data generation, model
training with dropout and early stopping techniques, and the use of pre-trained
models and data augmentation. The results showed that data augmentation did not
improve model accuracy compared to no augmentation, as both methods produced
the same accuracy of 90%. However, data augmentation improved model stability
and performance on validation data, while the model without augmentation
showed greater fluctuations in accuracy and stability. Model evaluation using
precision, recall, and f1-score also showed better results with augmentation,
proving the importance of this technique in improving the effectiveness of deep
learning models for eye disease detection.