AUTOMATIC DETECTION OF RETINAL LESIONS FOR SCREENING OF DIABETIC RETINOPATHY

Abstract

Diabetic Retinopathy (DR) is characterized by the progressive deterioration of retina with the appearance of different types of lesions that include microaneurysms, hemorrhages, exudates etc. Detection of these lesions plays significant role for early diagnosis of DR. Methods: To this aim, this paper proposes a novel and automated lesion detection scheme which consists of the four main steps: vessel extraction and optic disc removal, pre-processing, candidate lesion detection and post-processing. The optic disc and the blood vessels are suppressed first to facilitate further processing. Curvelet based edge enhancement is done to separate out the dark lesions from the poorly illuminated retinal background while the contrast between the bright lesions and the background is enhanced through an optimally designed wideband bandpass filter. The mutual information of the maximum matched filter response and the maximum Laplacian of Gaussian response are then jointly maximized. Differential Evolution algorithm is used to determine the optimal values for the parameters of the fuzzy functions that determine the thresholds of segmenting the candidate regions. Morphology based post-processing is finally applied to exclude the falsely detected candidate pixels. Results and Conclusions: Extensive simulations on different publicly available databases highlight an improved performance over the existing methods with an average accuracy of 97.71% and robustness in detecting the various types of DR lesions irrespective of their intrinsic properties.

Existing System

Points of interests (PoIs), ensemble of Gaussian mixture model and support vector machine (SVM)

Proposed System

The present work combines the use of matched filter (MF) with Laplacian of Gaussian (LoG) operation in an integrated framework to improve the lesion detection performance. MF produces strong responses not only to vessels but also to MAs and HEMs having Gaussian-like intensity profile. At sharp intensity transition of EXs, LoG response shows zero crossing about its center. Thus LoG filter, along with MF, is used to segment the transients like EXs and HEMs. The main contributions of the present work are as follows: Design of an optimal wideband bandpass filter (WBBF) for enhancement of bright lesions. • Differential Evolution (DE) based contrast enhancement to automatically set the gain and the bandwidth of the WBBF to make the system adaptive for different types of images based on their characteristics. • Maximization of mutual information (MI) of the maximum matched filter response (MFR) and the maximum Laplacian of Gaussian response (LoGR) in 2 dimensional (2D) feature space using DE which, to the best of our knowledge, is not explored earlier in lesion detection. • Significantly improved results over the existing methods when evaluated on a large set of images on different open access online databases.

Architecture

BLOCK DIAGRAM