Robust Vessel Segmentation in Fundus Images
Citations
1,053 citations
448 citations
429 citations
Cites methods from "Robust Vessel Segmentation in Fundu..."
...To the best of our knowledge, only unsupervised methods were previously tested on HRF [24], [55]....
[...]
317 citations
Cites methods from "Robust Vessel Segmentation in Fundu..."
...Four public datasets DRIVE [23], STARE [38], CHASE_ DB1 [39] and HRF [40] are used to evaluate the proposed...
[...]
...Finally, we apply the jointloss deep learning framework on the high-resolution fundus dataset HRF. Exemplar results1 are shown in the following sections....
[...]
...Four public datasets DRIVE [23], STARE [38], CHASE_ DB1 [39] and HRF [40] are used to evaluate the proposed segment-level loss and the joint-loss framework....
[...]
...HRF contains 15 images of healthy patients, 15 images of patients with diabetic retinopathy and 15 images of glaucomatous patients with a 60° FOV and all images have the same resolution of 3504 × 2336 pixels....
[...]
...After the data augmentation process, the numbers of training patches for the datasets DRIVE, STARE, CHASE_DB1, and HRF are 7869, 8677, 4789 and 4717 respectively....
[...]
280 citations
Cites background or methods from "Robust Vessel Segmentation in Fundu..."
...For images from DRIVE and STARE and CHASE_DB1 datasets, “p = 0.2,” which ensures the red regions to be highlighted in the vessel enhanced binary images....
[...]
...However, increasing the segmentation accuracy above 92% for abnormal retinal images with bright lesions (exudates and cotton wool spots), or red lesions (hemorrhages and microaneurysms), or variations in retinal illumination and contrast, while maintaining low computational complexity is a…...
[...]
References
28,073 citations
"Robust Vessel Segmentation in Fundu..." refers methods in this paper
...After the vessel enhancement is completed in each resolution level and the results are resized to original resolution, all of themare binarized by a thresholding algorithm proposed by Canny [31]....
[...]
18,095 citations
"Robust Vessel Segmentation in Fundu..." refers background in this paper
...The most known and commonly used ones are active contour-based methods, level-sets, and the socalled snakes [19]....
[...]
8,738 citations
"Robust Vessel Segmentation in Fundu..." refers methods in this paper
...Histogram stretching [28] and bilateral filtering [29] are applied to the green channel....
[...]
3,928 citations
"Robust Vessel Segmentation in Fundu..." refers background or methods in this paper
...Therefore, we optimized the parameters of both—the proposed and the Frangi—methods....
[...]
...[1] as a base for our own work, because it features some attractive properties....
[...]
...After that the vessel enhancement method is described to highlight the main differences to the Frangimethod....
[...]
...[1] in combination with a multiresolution framework to decrease the computational needs and to increase the sensitivity by using a hysteresis thresholding....
[...]
...[1] proposed, the method calculates the Hessian matrix and the given measures for increasing neighborhood sizes, until the neighborhood is bigger than the expected thickest vessel....
[...]
3,416 citations