The Information System Department at Universitas Pendidikan Ganesha is dedicated to advancing research that aligns with the Sustainable Development Goals (SDGs) set by the United Nations. One such impactful research initiative led by I Made Dendi Maysanjaya, a distinguished lecturer in the Information System Study Program, focuses on diabetic retinopathy, a critical eye condition prevalent among individuals with diabetes that can lead to vision impairment and blindness.
Diabetic retinopathy is a serious complication of diabetes characterized by damage to blood vessels and nerve fibers in the eye, primarily caused by the presence of exudates—deposits resulting from leaking blood vessels. Exudates are categorized into two types: hard exudate and soft exudate, with the latter being the focus of this research endeavor.
In pursuit of more effective diagnosis and treatment, this research leverages digital image processing techniques, specifically employing a deep learning-based method known as U-Net. The U-Net model is renowned for its proficiency in image segmentation tasks, crucial for accurately identifying and delineating soft exudate regions from non-soft exudate areas within retinal images.
The Segmentation Process:
The research methodology encompasses three key stages:
- Pre-processing: This initial stage enhances the quality of retinal images, ensuring optimal conditions for subsequent segmentation processes.
- Segmentation: Using the U-Net model, soft exudates are precisely differentiated from other elements in the images, aiding in the accurate identification of diabetic retinopathy markers.
- Evaluation: The performance of the segmentation approach is rigorously evaluated through metrics such as accuracy, sensitivity, and specificity, benchmarked against ground truth data for validation.
The outcomes of this study demonstrate the efficacy of the U-Net method in diabetic retinopathy diagnosis, with notable performance metrics including an average accuracy of 0.99586, sensitivity of 0.36203, and specificity of 0.99856. These results underscore the potential of deep learning techniques in revolutionizing medical imaging analysis and enhancing diagnostic capabilities, thereby contributing significantly to advancements in healthcare.
By supporting research endeavours such as this, the Information System Department reaffirms its commitment to the Sustainable Development Goals, particularly Goal 3: Good Health and Well-being. Through innovative technological interventions like deep learning-based image processing, the department strives to improve healthcare outcomes, promote early disease detection, and ultimately enhance the quality of life for individuals affected by diabetic retinopathy and other health challenges.
The collaborative efforts between researchers like I Made Dendi Maysanjaya and the Information System Department exemplify a proactive approach towards addressing global challenges outlined in the SDGs. Through pioneering research initiatives, informed by cutting-edge technologies and a commitment to societal impact, the department continues to make significant strides in advancing healthcare solutions and fostering sustainable development on a global scale.
