This project is a simple implementation of morphological operation based retinal vessel segmentation.
Sample Result: Red - true positive, Green - false positive, Blue - false negative
This step contains green channel extraction (because retinal image has highest contrast in green channel) and gaussian filtering (for denoising).
This step is for vessel enhancement which is based on the comparison between the input image and its opening and closing
- top-hat: the difference between the original image and its opening
The top-hat transform is applied so that background areas are set to zero and vessels are enhanced. The sum of several top-hat transforms is later AND with the eroded mask to get rid of the bright boundary.
In this code, the top-hat is implemented by min-max filter, i.e. erosion is min filter, dilation is max filter
A linear normalization is then performed to reduce the effect of variance of image intensities by using the formula described in Wikipedia as:
After trying several threshold values (50, 60, 70, etc), 60 is set to get comparatively optimal segmentation performance.
This project is developed in VisionX using C and shell script.
[1] Rossant, Florence, et al. "A morphological approach for vessel segmentation in eye fundus images, with quantitative evaluation." Journal of Medical Imaging and Health Informatics 1.1 (2011): 42-49.
[2] Walter T, Klein JC. Segmentation of color fundus images of the human retina: Detection of the optic disc and the vascular tree using morphological techniques”. International Symposium on Medical Data Analysis 2001 Oct 8 (pp. 282-287).
[3] Zana F, Klein JC. “Segmentation of vessel-like patterns using mathematical morphology and curvature evaluation”. IEEE Trans. Image Processing. 2001 Jul;10(7):1010-9.