Hierarchical retinal blood vessel segmentation based on feature and ensemble learning
Introduction
Retinal blood vessel segmentation has been widely used in various scenarios. For example, change of the retinal blood vessel appearance is an important indicator for various ophthalmologic and cardiovascular diseases, such as diabetes, hypertension, and arteriosclerosis [1], therefore, automatic segmentation and analysis of the retinal vasculature play an extremely vital role in the implementation of screening programs for diabetic retinopathy, the evaluation of retinopathy of prematurity, foveal avascular region detection, arteriolar narrowing detection, the diagnosis of cardiovascular diseases and hypertension, and computer-assisted laser surgery [2]. Moreover, the generation of retinal maps and detection of branch points have been utilized for temporal or multimodal image registration, retinal image mosaic synthesis, optic disc identification, fovea localization and biometric identification [2].
Both manual delineation and automatic algorithms have been used in retinal vessel segmentation. However, they have not gained wide acceptance due to several challenges. Manual delineation is skill demanding, tedious, time-consuming, and infeasible if given a large volume of fundus image databases. Accuracy of the automatic segmentation algorithms (to be reviewed in Section 2) is limited due to low blood vessel contrast, irregular shaped bright and dark lesions (in the form of hemorrhages, exudates, drusen and the optic disc boundary), intricate vessel topology (including vessel crossing and branching, as well as variation of vessel diameter and vessel grey levels) and nonuniform illumination of images as well as image deformation of scaling, skewing and other distortions.
In this paper, we present the hybrid method based upon convolutional neural network (CNN) [3] and ensemble random forests (RFs) [4] for automatic retinal blood vessel segmentation. We first employed a set of preprocessing steps to correct the nonuniform illumination of retinal images and to improve vessel contrast. We then used CNN to extract a set of hierarchical features which are not only invariant to image translation, scaling, skewing and other distortions, but also contain image based multi-scale information of the geometric structure of retina. We finally trained ensemble RFs to obtain a vessel classifier. The whole pipeline of the proposed method is trainable and automatic. Moreover, our method can effectively deal with the challenges of retinal vessel segmentation, as shown by our evaluations conducted using two publicly available databases (the DRIVE [5] and STARE [1]) and comparisons with state-of-the-art. Experimental results show that our approach is competitive with state-of-the-art by achieving sensitivity/specificity/accuracy/AUC values of 0.8173/0.9733/0.9767/0.9475 for the DRIVE database and of 0.8104/0.9791/0.9813/0.9751 for the STARE database. In contrast, previous methods of Fraz [6] produced values of 0.7406/0.9807/0.9480/0.9747 for the DRIVE database and of 0.7548/0.9763/0.9534/0.9768 for the STARE database.
The rest of this paper is organized as follows. Section 2 gives a review of related work reported in the latest literature. Section 3 provides the description of CNN and RF. Section 4 gives a detailed description of the proposed method. Section 5 evaluates the proposed method. Further, the concluding remarks are included in Section 6.
Section snippets
Related work
There are a number of methods available in the literature for retinal blood vessel segmentation, as reviewed by Fraz et al. [2]. These methods can be broadly divided into two categories: unsupervised and supervised [2], [3].
Methodology
The proposed method leverages trainable hierarchical features extracted using the CNN algorithm [3] and performs classification based on the random forest (RF) technique [4]. In this section, we will briefly review the CNN and RF techniques.
Proposed method
In the proposed method, CNN performs as a trainable hierarchical feature extractor and ensemble RFs work as a classifier. It is noteworthy that features utilized in the proposed method learned from not only the last layer output of CNN but also the intermediate layers outputs. Features learned from the same layer of CNN are fed into one RF classifier. Finally, winner-takes-all is employed to ensemble the predictions of each RF. For training procedure, raw pixel values from a square sub-window
Materials
Similar to most of the retinal vessel segmentation methods, the proposed method is evaluated on two well established public databases: the DRIVE database and the STARE database.
Conclusion
In this paper, we proposed a hierarchical retinal blood vessel segmentation method based on feature and ensemble learning. The proposed method has several unique characteristics. First, our features are extracted using not only the last layer output but also the intermediate output of CNN and therefore contain multiscale information of the geometric structure of the retina. Second, we are the first to introduce random forest into retinal blood vessel segmentation, and employ winner-takes-all as
Acknowledgments
The authors would particularly like to thank the anonymous reviewers for their helpful suggestions. This work is supported by NSFC Joint Fund with Guangdong under Key Project U1201258, Program for New Century Excellent Talents in University of Ministry of Education of China (NCET-11–0315) and Shandong Natural Science Funds for Distinguished Young Scholar under Grant no. JQ201316.
Shuangling Wang received her BS in 2011 from the Shandong University, Weihai. Now she is studying at the Shandong University for a Master degree in computer application technology. Her main research interests are machine learning and biomedical image processing.
References (36)
- et al.
Blood vessel segmentation methodologies in retinal images–A survey
Comput. Methods Progr. Biomed.
(2012) - et al.
An approach to localize the retinal blood vessels using bit planes and centerline detection
Comput. Methods Progr. Biomed.
(2012) - et al.
Detection of neuron membranes in electron microscopy images using a serial neural network architecture
Med. Image Anal.
(2010) - et al.
Retinal vessel segmentation using multiwavelet kernels and multiscale hierarchical decomposition
Pattern Recognition
(2013) - et al.
Segmentation of retinal blood vessels using the radial projection and semi-supervised approach
Pattern Recognit.
(2011) - et al.
Locating blood vessels in retinal images by piecewise threshold probing of a matched filter response
IEEE Trans. Med. Imaging
(2000) - et al.
Convolutional networks for images, speech, and time series
The Handbook of Brain Theory and Neural Networks
(1995) Random forests
Mach. Learn.
(2001)- et al.
Ridge-based vessel segmentation in color images of the retina
IEEE Trans. Med. Imaging
(2004) - et al.
An ensemble classification-based approach applied to retinal blood vessel segmentation
IEEE Trans. Biomed. Eng.
(2012)
Detection and measurement of retinal vessels in fundus images using amplitude modified second-order Gaussian filter
IEEE Trans. Biomed. Eng.
Segmentation of vessel-like patterns using mathematical morphology and curvature evaluation
IEEE Trans. Image Process.
Segmentation of retinal blood vessels by combining the detection of centerlines and morphological reconstruction
IEEE Trans. Med. Imaging
Vessel extraction in medical images by wave-propagation and traceback
IEEE Trans. Med. Imaging
Multiscale vessel enhancement filtering,
Medical Image Computing and Computer-Assisted Interventation—MICCAI׳98
Analysis of retinal vasculature using a multiresolution hermite model
IEEE Trans. Med. Imaging
A novel vessel segmentation algorithm for pathological retina images based on the divergence of vector fields
IEEE Trans. Med. Imaging
General retinal vessel segmentation using regularization-based multiconcavity modeling
IEEE Trans. Med. Imaging
Cited by (0)
Shuangling Wang received her BS in 2011 from the Shandong University, Weihai. Now she is studying at the Shandong University for a Master degree in computer application technology. Her main research interests are machine learning and biomedical image processing.
Yilong Yin is the Director of MLA Lab and a Professor of the Shandong University. He received his Ph.D. degree in 2000 from the Jilin University. From 2000 to 2002, he worked as a post-doctoral fellow in the Department of Electronic Science and Engineering, Nanjing University. His research interests include machine learning, data mining, and biometrics.
Guibao Cao received her BSEE in 2011 from the Harbin University of Commerce. Now he is studying at the Shandong University for a Master degree in computer application technology. His main research interests are pattern recognition, machine learning, and image processing.
Benzheng Wei received B.Sc., M.Sc. degrees in Computer Science and Technology from Shandong Institute of Light Industry and Shandong University, China in 2000 and 2007, respectively. Form 2010, he has been an Associate Professor in the College of Science and Technology, Shandong University of Traditional Chinese Medicine. His research interests focus on Medical image processing, Pattern analysis and Medical information processing.
Yuanjie Zheng is a senior research investigator in the Perelman School of Medicine at the University of Pennsylvania and the primary contact of Image Analysis Core at the Penn Vision Research Center. He received his Ph.D. degree in Pattern Recognition and Intelligent Systems from Shanghai Jiao Tong University of China in 2006. His current research interests are in computer vision, medical image analysis and clinical study.
Gongping Yang received his Ph.D. degree in computer software and theory from Shandong University, China, in 2007. Now he is a professor in School of Computer Science and Technology, Shandong University. His research interests are biometrics, medical image processing, and so forth.