Author/Authors :
abd aziz, nurhakimah universiti teknologi mara - school of electrical engineering, college of engineering, Shah Alam, Malaysia , sulaiman, mohd azman hanif universiti teknologi mara - school of electrical engineering, college of engineering, Shah Alam, Malaysia , mohd yassin, ahmad ihsan universiti teknologi mara - school of electrical engineering, college of engineering, Shah Alam, Malaysia , megat ali, megat syahirul amin universiti teknologi mara - school of electrical engineering, college of engineering, Shah Alam, Malaysia , abu hassan, hasliza universiti selangor (unisel) - faculty of electrical engineering, Bestari Jaya, Malaysia , shafie, suraiya m. prince court medical centre (pcmc), Kuala Lumpur, Malaysia , zabidi, azlee universiti malaysia pahang (ump) - faculty of computing, Pahang, Malaysia , eskandari, farzad allameh tabatabai university - department of mathematical science and computer, Tehran, Iran
Abstract :
In recent years, the lesions detection in fundus image become popular area of research in machine learning. The detection of symptoms in fundus image is typically used in diseases that related to eyes such as diabetic retinopathy where the main symptom is exudates. Symptom detection in fundus image depends on many factor. The common factors are varying contrast condition and the large size of the fundus image that will affect the training process for object detection. Furthermore, color similarity of the features in fundus image and the symptoms also one of the factor, for example the similarity between optics disc and exudates. In this paper, we discuss the different preprocessing stage in order to improve the quality of fundus image to mark the optic disc location for detection of optic disc in future work. We have used several datasets namely Kaggle, DIARETDB1 and DRIMDB datasets in this study. The results that we have achieved in SSIM value, clearly shows that the preprocessing was able to increase the image quality.
Keywords :
Contrast Variation , Diabetic Retinopathy , Fundus Image , Optic Disc , Preprocessing
