Application of gray level run length matrices features extraction for diabetic retinopathy detection based on artificial neural network

• Main Authors: Wardani, Bestia Kumala, Belinda, Nathania Earlene, Rulaningtyas, Riries, Purwanti, Endah
• Format: Conference Proceeding
• Language: English
• Subjects: Artificial neural networks; Back propagation; Back propagation networks; Blindness; Diabetes; Diabetes mellitus; Diabetic retinopathy; Eye (anatomy); Feature extraction; Neural networks; Nonuniformity; Texture; Training
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0035182

Diabetic Retinopathy (DR) is Diabetes Mellitus microvascular complication, which attacks the eyes and can cause blindness. The DR examination currently is done with the fundus photography technique. The image produced by the technique is studied by the ophthalmologist manually. This research aimed to develop a computer aided diagnosis in the DR detection system using Gray Level Run Length Matrices (GLRLM) features extraction and Artificial Neural Network. This DR detection was done by Gray Level Run Length Metrices (GLRLM) texture features extraction method. Five texture features used were: Short Run Emphasis (SRE), Long Run Emphasis (LRE), Gray Level Non-Uniformity (GLN), Run Length Non-Uniformity (RLN), and Run Perecentage (RP). The five features were used as Backpropagation Artificial Neural Network input. The parameters are varied in the Backpropagation Artificial Neural Network training phase. The data used in this research were 110 retinal fundus images consists of 20 normal retinal fundus images, 45 NPDR retinal fundus images, and 45 PDR retinal fundus images. The highest accuracy result gained in the training phase is 97.14% and testing phase is 92.5%. This detection program produces a sensitivity value of 93.33% and a specificity value of 90%, therefore, it can be used as an initial indicator to detect DR.