Unified adaptive framework for contrast enhancement of blood vessels

Abstract

Information about blood vessel structures influences a lot of diseases in the medical realm. Therefore, for proper localization of blood vessels, its contrast should be enhanced properly. Since the blood vessels from all the medical angio-images have almost similar properties, a unified approach for the contrast enhancement of blood vessel structures is very useful. This paper aims to enhance the contrast of the blood vessels as well as the overall contrast of all the medical angio-images. In the proposed method, initially, the vessel probability map is extracted using hessian eigenanalysis. From the map, vessel edges and textures are derived and summed at every pixel location to frame a unique fractional differential function. The resulting fractional value from the function gives out the most optimal fractional order that can be adjusted to improve the contrast of blood vessels by convolving the image using Grunwald-Letnikov (G-L) fractional differential kernel. The vessel enhanced image is Gaussian fitted and contrast stretched to get overall contrast enhancement. This method of enhancement, when applied to medical angio-images such as the retinal fundus, Computerised Tomography (CT), Coronary Angiography (CA) and Digital Subtraction Angiography (DSA), has shown improved performance validated by the performance metrics.