Automated Detection and Analysis of Retinal Blood Vessels in Retinopathy of Prematurity Using Image Processing Techniques

Retinal disorders pose a significant threat to human vision, with delayed or untreated cases leading to irreversible vision loss, including total blindness. Among these disorders, retinopathy of prematurity (ROP) stands out as a leading cause of blindness in preterm newborns due to abnormal retinal blood vessel development. While efforts have been made to diagnose retinal disorders early, most existing approaches have fallen short, particularly in addressing ROP. This study centers on the crucial task of retinal blood vessel segmentation in ROP fundus images, coupled with precise measurements of vessel length, breadth, and tortuosity. These metrics serve as valuable indicators for early ROP diagnosis and risk assessment, aiding ophthalmologists in preventing infant blindness. Our approach begins by enhancing ROP image quality, addressing issues of poor contrast, noise, and uneven lighting. We then employ morphological operations, lookup tables, and geodesic distance transformation techniques to delineate vessel characteristics accurately. Experimental results demonstrate the superiority of our method in both blood vessel segmentation and morphological measurements compared to existing techniques. This research contributes a robust and effective tool for ROP diagnosis, offering promise in mitigating the threat of infant blindness and advancing the field of retinal image analysis.