TBME-00468-2010-R1 - IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING 1 A Successive Clutter-Rejection based Approach for Early Detection of Diabetic Retinopathy Keerthi Ram, Gopal Datt Joshi, Member, IEEE, and Jayanthi Sivaswamy, Member, IEEE Abstract—The presence of microaneurysms (MAs) is usually an early sign of diabetic retinopathy(DR) and their automatic detection from color retinal images is of clinical interest. In this paper, we present a new approach for automatic MA detection from digital colour fundus images. We formulate MA detection as a problem of target detection from clutter, where the probability of occurrence of target is considerably smaller compared to the clutter. A successive rejection-based strategy is proposed to progressively lower the number of clutter responses. The processing stages are designed to reject specific classes of clutter while passing majority of true MAs, using a set of specialized features. The true positives that remain after the final rejector are assigned a score which is based on its similarity to a true MA. Results of extensive evaluation of the proposed approach on three different retinal image datasets is reported, and are used to highlight the promise in the presented strategy. Index Terms—Diabetic retinopathy, Microaneurysm, Clutter- rejection, Retinal image I. I NTRODUCTION D IABETIC Retinopathy (DR) is a major public health issue since it can lead to blindness in patients with diabetes. Microaneurysms (MAs) are usually the first clinical symptom of DR. They are swellings of capillaries caused by a weakening of the vessel wall [1]. Their sizes range from 10μm to 125μm [2]. In the clinical scenario, experts rely either on direct manual examination or fluorescein fundus angiog- raphy where MAs appear with high contrast as bright white spots. Given the high cost and the cumbersome requirement of intravenous injection of a dye for this type of imaging, interest in the recent past has been on detecting MAs from a colour fundus/retinal image(CFI). In CFIs, MAs appear as tiny, reddish isolated dots. Automatic detection of MAs from digital CFIs can play an important role in DR screening at a large scale [3][4]. It can significantly reduce the workload of the ophthalmologists and the health costs in the DR screening [3]. Early published work attempted to address the problem of MA detection in fluorescein angiogram images of the retina [5][6][7][8][9]. Lay et al., [5] presented the first MA detection method for angiograms. In this method, MA candidates were obtained using top-hat transformation which eliminates the Manuscript received May 25, 2010; revised Aug 12, 2010. This work was supported partly by the Department of Science and Technology, Govt. of India, under Grant SR/S3/EECE/17/2005. All authors are with the Centre for Visual Information Technol- ogy, IIIT Hyderabad, India, 500032. e-mail: keerthiram@research.iiit.ac.in, gopal@research.iiit.ac.in and jsivaswamy@iiit.ac.in Copyright (c) 2010 IEEE. Personal use of this material is permitted. However, permission to use this material for any other purposes must be obtained from the IEEE by sending an email to pubs-permissions@ieee.org vasculature structure from the image leaving possible MA candidates untouched. Spencer et al., [7] presented a shade correction technique and a candidate detection method using matched filtering. However, potential mortality associated with the intravenous use of fluorescein [10][4] prohibits the appli- cation of this technique for large-scale screening purposes. Instead, colour fundus imaging has emerged as a preferred modality due to its non-invasive nature [10]. Extensive clinical studies show the effectiveness of CFI for large-scale DR screening [3]. Existing methods for MA detection generally consist of two- stages where, the first stage is aimed at obtaining potential MA candidates while the second stage is used to assign MA or non-MA category to the candidate using features computed around the candidate location. The main processing components include 1) pre-processing; selection of candidate MA and 2) feature extraction; classification. The focus of the early methods has been on pre-processing and candidates selection steps. Later methods focus more on designing new sets of features and choosing of classifiers. Recently published work have re-examined the individual processing components and presented improvements on certain aspects. Numerous algorithms have been proposed to detect early signs of DR (MAs) from CFI. The first such method was presented by Oien et al. [11]. The pre-processing used here is similar to the approach used by [5]. In later methods, a rule-based classification was added to the processing pipeline [6][8][12][13]. Usher et al.,[14] employed a neural network based classification on the candidate regions obtained using recursive region growing and adaptive intensity thresholding. Huang et al.,[2] presented a local adaptive approach to extract candidates, where multiple subregions of each image were automatically analyzed to adapt to local intensity varia- tion and properties. Niemeijer et al.,[4] presented a supervised, pixel classification technique to extract red lesions to get MA candidates. A large set of features was added to the original feature set used in [6]. A knn classifier was used for MA recognition. Fleming et al.,[1] presented a local image contrast normalization technique to get more discriminative features for MA. A vessel-free region is obtained around each detected candidate using watershed segmentation which is then used to enhance the contrast of candidate. A parametric model of a paraboloid is used for the MA and fitted on a set of pixels obtained by applying region growing on the candidate location. The model parameters are used to derive a new set of features for the candidate and finally classified using a knn classifier. Walter et al.,[15] used a morphological (diameter) closing technique for detecting candidates. A supervised density-based