Retinal Vessel Segmentation Using Modify Matching Filter and Hessian Matrix

Blood vessel segmentation is a difficult assignment in medical image processing for illness diagnosis. It analyzes vessels critical in automatic retinal vessel extraction using various methodologies, strategies, and algorithms to forecast diseases such as laryngology, neurosurgery, and ophthalmology. Several diseases are common causes of early vision loss and blindness, such as diabetic retinopathy, glaucoma, age-related macular edema, retinal detachment, retinal tears, retinal vascular occlusion, and macular holes. Therefore, early detection of such diseases is mainly important because it allows timely treatment to avoid loss of life. To overcome this problem, a novel approach is introduced to segment the vessels based on the stretching of the histogram, as well as a modified active contour method that uses a separate wavelet transform for energy minimization. To avoid destroying the evolutionary stability, we employed a process for optimizing the formula whose task is to retain the contour on the borders of the twenty images on the DRIVE dataset. This paper presents techniques for segmenting retinal vessels that go through three stages: stage before processing, stage during processing, and stage after processing. The proposed algorithm's effectiveness is tested with other studies and manually set data. The result of this method is better than other works, and values were achieved (0.9416, 0.661158, and 0.983259) for accuracy (ACC), sensitivity (SE), and specificity (SP), respectively. This technique can also be modified and applied to help clinicians diagnose retinal diseases using image-processing techniques.