| Digital fundus image has become an advantageous screening tool of many eye diseases in recent years, because it can be acquired non-invasively and observed intuitively. Glaucoma and Diabetic Retinopathy(DR) are two kinds of three human blinding eye diseases, and they can lead to irreversible diminution of vision if not be treated timely. Therefore, the early discovery becomes one of the most important measures for reducing blindness. In the clinical, glaucoma and DR screening need doctors to review fundus images by personal experience, which is subjective, fatiguing, time-consuming, and imefficient. For this problem, this study combined medical image processing technology with expertise knowledge effectively to segment optic cup which is closely related with glaucoma and detect Hard Exudates(HE) in DR early stage, which lays the foundation for glaucoma and DR computer-aided screening system. The main research contents of this paper are as below:1. An automatic segmentation algorithm of optic cup was implemented. First of all, morphological operations were exploited to perform the contrast enhancement of green channel image; then, the blood vessels were detected and filled by an improved Bertalmio–Sapiro–Caselles–Ballester(BSCB) model; finally, the cup was segmented based on the Local Chart-Vest(LCV) model approach. Totally 94 images(62 normal images and 32 glaucoma images) in the public database were used for the algorithm evaluation. The F-score and boundary distance of the segmented cup areas by the proposed method were better than the results obtained by other methods. The average Cup-to-Disc Ratio(CDR) of normal and glaucoma images measured by proposed algorithm were 0.4369±0.1193, 0.1193±0.7156, respectively, which were similar to the doctor's measurement.The experimental results showed that the proposed optic cup segmentation algorithm gets high accuracy and good feasibility.2. A detection algorithm of HE was realized. Algorithm in this paper used the characteristics of brightness and edge in HE, and proposed a new automatic HE detection method based on an improved Canny edge detection algorithm combined with morphological remodeling method. This method can be divided into four steps. The first step is the image preprocessing, mainly including the selection of RGB channels and image contrast enhancement based on morphology; the second step is the elimination of key structures in retinal image. To avoid interference to the HE detection, the vessel segmentation method based on Gabor filter is used to eliminate the influence of blood vessel edge, and the proposed optic cup segmentation method is used in red channel to eliminate the optic disk and its edge; the third step is to extract the HE by using the improved Canny edge detection operator combined with morphological reconstruction method; the forth step is image post-prosessing based on morphology, eliminating the false positive area in image edges. We tested 40 images in the public database(35 images with HE lesions, 5 normal images). Its lesion based sensitivity and positive predictive value were 93.18%, 79.26%, respectively. Its image based sensitivity, specificity and accuracy were 97.14%, 80.00% and 95.00%, respectively. We compared the above evaluation indexes with other methods, and the results proved the feasibility of the algorithm.In short, this study realized the automatic segmentation of optic cup and the automatic detection of HE in fundus image, and showed the feasibility of above algorithms by experiments, laying the foundation for glaucoma and DR computeraided screening system. |
PDF Full Text Request