Abstract The localization of optic disc (OD) is very important for computer-aided diagnosis of the ophthalmology diseases with fundus images. In this paper, a method of OD localization based on regional proposal strategy was proposed. First, the fundus image was mapped from the pixel domain to the feature domain, and candidate regions of OD were generated by using the regional proposal strategy in the obtained feature maps. Next, the candidate regions were sampled according to the certain criteria, and a fully connected layer was constructed to perform deep feature extraction. The location refinement of the candidate region was achieved by using the constraint of the loss function. At last, OD visibility was judged by filtering of the confidence threshold. If the OD was visible, the center of the candidate region with the highest degree of confidence was regarded as OD coordinate of the fundus image. The correct position of OD was obtained. Experiments were conducted in three public fundus image databases (DRIVE (40 images), MESSIDOR (1200 images) and STARE (400 images)). Testing results were 100%, 99.9% and 98.8%. Experimental results showed that the proposed method could reach the OD localization fast, accurately and robust, which was superior to existing OD localization methods. The pre-judgment of OD visibility was more consistent with the requirements of practical application. The proposed method was expected to contribute to the diagnosis of fundus diseases.
Tang Yiping,Wang Liran,He Xia, et al. Optic Disc Localization Based on Regional Proposal Strategy[J]. Chinese Journal of Biomedical Engineering, 2019, 38(1): 9-17.
[1] Lu S, Lim JH. Automatic optic disc detection from retinal images by a line operator[J]. IEEE Transactions on Biomedical Engineering, 2011, 58(1): 88-94. [2] 邹北骥, 张思剑, 朱承璋. 彩色眼底图像视盘自动定位与分割[J]. 光学精密工程, 2015, 23(4):1187-1195. [3] Hoover A, Goldbaum M. Locating the optic nerve in a retinal image using the fuzzy convergence of the blood vessels[J]. IEEE Transactions on Medical Imaging, 2003, 22(8):951-958. [4] Welfer D, Scharcanski J, Kitamura CM, et al. Segmentation of the optic disk in color eye fundus images using an adaptive morphological approach[J]. Computers in Biology & Medicine, 2010, 40(2):124-137. [5] Mahfouz AE, Fahmy AS. Fast localization of the optic disc using projection of image features[J]. IEEE Transactions on Image Processing, 2010, 19(12):3285-3289. [6] Youssif AR, Ghalwash AZ, Ghoneim AR. Optic disc detection from normalized digital fundus images by means of a vessels′ direction matched filter[J]. IEEE Transactions on Medical Imaging, 2008, 27(1): 11-18. [7] Liu W, Anguelov D, Erhan D, et al. SSD: Single shot multibox detector[C] //European Conference on Computer Vision. Amsterdam: Springer International Publishing, 2016: 21－37. [8] Jonathan L, Evan S, Trevor D. Fully convolutional networks for semantic segmentation[C] //IEEE Conference on Computer Vision and Pattern Recognition. Boston: IEEE, 2015: 3431-3440. [9] Esteva A, Kuprel B, Novoa RA, et al. Corrigendum: Dermatologist-level classification of skin cancer with deep neural networks [J]. Nature, 2017, 542(7639):115-118. [10] 张贵英, 张先杰. 基于深度学习的视盘自动检测[J]. 贵州师范学院学报, 2017, 33(3):27-32. [11] Girshick R, Donahue J, Darrell T, et al. Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation[C] //IEEE Conference on Computer Vision and Pattern Recognition. Columbus: IEEE, 2014: 580-587. [12] Girshick R. Fast R-CNN [C] //International Conference on Computer Vision. Boston: IEEE, 2015: 1440-1448. [13] Ren SQ, He KM, Girshick R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[J]//IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017,39:1137-1149. [14] Staal J, Abràmoff MD, Niemeijer M, et al. Ridge-based vessel segmentation in color images of the retina [J]. IEEE Transactions on Medical Imaging, 2004, 23(4):501-509. [15] Aquino A, Gegundez-Arias ME, Marin D. Detecting the optic disc boundary in digital fundus images using morphological, edge detection, and feature extraction techniques[J]. IEEE Transactions on Medical Imaging, 2010, 29(11):1860-1869. [16] Hoover A, Kouznetsova V, Goldbaum M. Locating blood vessels in retinal images by piecewise threshold probing of a matched filter response[J]. IEEE Transactions on Medical Imaging, 2000, 19(3): 203-210. [17] 吴慧, 陈再良, 欧阳平波,等. 基于置信度计算的快速眼底图像视盘定位[J]. 计算机辅助设计与图形学学报, 2017, 29(6):984-991. [18] 张东波, 易瑶, 赵圆圆. 基于投影的视网膜眼底图像视盘检测方法[J]. 中国生物医学工程学报, 2013, 32(4):477-483. [19] Yu H, Barriga ES, Agurto C, et al. Fast localization and segmentation of optic disk in retinal images using directional matched filtering and level sets[J]. IEEE Trans Inf Technol Biomed, 2012, 16(4):644-657. [20] Harangi B, Hajdu A. Detection of the optic disc in fundus images by combining probability models[J]. Computers in Biology & Medicine, 2015, 65:10-24. [21] 肖志涛, 邵一婷, 张芳,等. 基于眼底结构特征的彩色眼底图像视盘定位方法[J]. 中国生物医学工程学报, 2016, 35(3):257-263. [22] 柯鑫, 江威, 朱江兵. 基于视觉注意机制的眼底图像视盘快速定位与分割[J]. 科学技术与工程, 2015, 15(35):47-53. [23] 赵圆圆, 张东波, 刘茂. 采用局部搜索的快速视盘检测方法[J]. 光电工程, 2014, 41(3):28-34. [24] 郑绍华, 陈健, 潘林,等. 基于定向局部对比度的眼底图像视盘检测方法[J]. 中国生物医学工程学报, 2014, 33(3):289-296. [25] Foracchia M, Grisan E, Ruggeri A. Detection of optic disc in retinal images by means of a geometrical model of vessel structure[J]. IEEE Transactions on Medical Imaging, 2004, 23(10):1189-1195.
