Mata adalah salah satu panca indera manusia yang kurang mendapat perhatian meskipun fungsinya yang sangat vital bagi manusia. hal ini menyebabkan banyak penyakit dan gangguan yang menyerang mata kita. Bahkan, beberapa penyakit dapat dilihat dari mata antara lain tekanan darah tinggi, kanker dan diabetes. Salah satu gejala diabetes adalah perdarahan ringan pada retina, adalah gejala Diabetic Retinopathy (DR). Jika tidak diobati, kondisi itu dapat berakhir dengan kebutaan. Segmentasi pembuluh darah pada citra fundus mata dapat membantu analisa terhadap perubahan pembuluh darah di retina. Pada penelitian Tugas Akhir ini, segmentasi pembuluh darah retina dimulai dengan tahap preprocessing, dan segmentasi. Tahap pertama adalah preprocessing, yaitu mengambil citra green channel,dan memperbaiki kontras citra dengan Contrast Limited Adaptive Histogram Equalization (CLAHE). Tahap kedua adalah segmentasi dengan First-Order Derivative of Gaussian, dilanjutkan dengan Morfologi Top Hat. Dilakukan perbaikan dengan proses noise removal, overlapping mask, operasi morfologi dilasi dan field of view removal. Hasil dari proses tersebut dibandingkan dengan citra groundtruth untuk menghitung tingkat akurasi. Pengujian dilakukan pada database DRIVE sebanyak 20 citra. Citra hasil segmentasi dibandingkan dengan citra groundtruth. Akurasi yang didapat dari pengujian tersebut sebesar 88,79%. Citra hasil segmentasi kurang baik sehingga metode First-Order Derivative of Gaussian mampu melakukan segmentasi dengan cukup baik Kata kunci : Segmentasi, First-Order Derivative of Gaussian, Morfologi Tophat, Fundus Mata.

ABSTRACT Eye is one of human five senses which gets less attention although it has a vital function for human. Thus, there are many diseases and nuisances that attacking our eye. Moreover, several diseases which can be seen through the eye such as, high blood pressure, cancer and diabetes. One of the diabetes symptoms is light bleeding on retina that becomes the symptom of Diabetic Retinopathy (DR). If it does not treated, this condition can cause blindness. A segmentation of blood vessels toward the image of eye fundus can help to analyze the alteration of blood vessels in retina.In this study, the segmentation of retina’s blood vessels is begun with preprocessing stage, and segmentation. The first stage is preprocessing, that is take the image of green channel and improve the image contrast by Contrast Limited Adaptive Histogram Equalization (CLAHE). The second stage is segmentation by First-Order Derivative of Gaussian, and continued by Top Hat Morphology. The improvement is done by the process of noise removal, overlapping mask, operation of dilation morphology and field of view removal. The result of those processes is compared with the image of groundtruth to calculate the accuracy. The test is done at DRIVE database at the rate of 20 images. The image of segmentation result is compared with the image of groundtruth. The accuracy which is obtained through those tests is 88, 79%. The image of segmentation result is unfavourable, so that the method of First-Order Derivative of Gaussian can do the segmentation fairly well. Keywords: segmentation, First-Order DerivativeGaussian, Top hat Morphology, eye fundus.

Baca Jurnal Baca Jurnal Baca Jurnal Baca Jurnal Baca Jurnal Baca Jurnal Baca Jurnal Baca Jurnal