An Automatic Detection of Retinal Lesions for Screening of Diabetic Retinopathy

The identification of retinal lesions in fundus pictures is necessary for an automatic telemedicine system for computer-assisted screening and grading of diabetic retinopathy. This study defines and validates an entirely new method for automatically identifying each microaneurysm and haemorrhage in colour fundus images. The most significant contribution has come from a new set of form options known as “dynamic form options,” which do not require precise segmentation of the regions to be classified. These options allow you to differentiate between lesions and vessel segments while also displaying how the morphology changed during image flooding. Six databases are used in the technique, four of which are open to the public. It is valid for each lesion on each image. It shows strength in image quality, resolution, and acquisition system variability. The method comes in fourth place with a FROC score of 0.420, according to the Retinopathy Online Challenge data. The planned method successfully achieves a section of 0.899 for detecting photos with diabetic retinopathy. Messidor data show that traditional methods underperform human consultants’ scores and love the score of traditional methods.