Hard Exudates Detection Method Based on Background-Estimation

Xiao, Zhitao; Li, Feng; Geng, Lei; Zhang, Fang; Wu, Jun; Zhang, Xinpeng; Su, Long; Shan, Chunyan; Yang, Zhenjie; Sun, Yu; Yu, Xiao; Du, Wei-Qiang

doi:10.1007/978-3-319-21963-9_33

Cited by 7 publications

(3 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Table . 5 indicates that our proposed algorithms achieved higher accuracy values than the existing methods [12], [51]- [56], and [57] in literature review. The performance of our proposed algorithms, compared to [54], has a nearby accuracy values of sensitivity at 97.00% and of specificity at 96.00%, respectively.…”

Section: Comparison Of the Proposed Methodsmentioning

confidence: 85%

Exudates Detection Using Morphology Mean Shift Algorithm in Retinal Images

Wisaeng

Sa-ngiamvibool

2019

IEEE Access

View full text Add to dashboard Cite

Exudates are a serious complication causing blindness in diabetic retinopathy patients. The main objective of this paper is to develop a novel method to detect exudates lesions in color retinal images by using a morphology mean shift algorithm. The proposed method start with a normalization of the retinal image, contrast enhancement, noise removal, and the localization of the OD. Then, a coarse segmentation method by using mean shift provides a set of exudates and non-exudates candidates. Finally, a classification using the mathematical morphology algorithm (MMA) procedure is applied in order to keep only exudates pixels. The optimal value parameters of the MMA will facilitate an increase of the accuracy results from the solely MSA method by 13.10%. Based on a comparison between the results and ground truth images, the proposed method obtained an average sensitivity, specificity, and accuracy of detecting exudates as 98.40%, 98.13%, and 98.35%, respectively.

show abstract

Section: Comparison Of the Proposed Methodsmentioning

confidence: 85%

Exudates Detection Using Morphology Mean Shift Algorithm in Retinal Images

Wisaeng

Sa-ngiamvibool

2019

IEEE Access

View full text Add to dashboard Cite

show abstract

“…(v) Classification- Garcia et al (2009a, b) proposed and evaluated the use global and adaptive thresholding, selection of features by logistic regression followed by classification using RBF neural Network for 117 images collected from a screening programs with variable quality, color, and brightness and reported a sensitivity of 100%, but specificity falls to 70.4%. Xiao et al (2015) extracted exudates using edge detection and then classified by SVM classifier using features like shape, statics, and phase. An AdaBoost classifier was proposed by Roychowdhury et al (2014) and reported a sensitivity and specificity of 98% and 74.2%, respectively.…”

Section: Exudate Detectionmentioning

confidence: 99%

Automated macula proximity diagnosis for early finding of diabetic macular edema

Patil

2020

Res. Biomed. Eng.

View full text Add to dashboard Cite

Purpose Diabetic retinopathy (DR) is non-recoverable in nature. One of the advanced sight threatening condition in diabetic patient is defined in terms of diabetic macular edema (DME), where the macula gets deposited with fluid rich in proteins called exudates. It is indisputably required to find and treat occurrence of exudates near macula in time, to avoid further complications of retina and vision loss at later stage. However, presence of various dark and bright lesions awakes the need of reliable macula and exudate detection process. Proposed work intends to present a robust, scale, and rotation invariant tool for early diagnosis of DME. Methods Optic disc (OD) diameter is one of the important parameters; it is always proportional to the size of the retina; an adaptive method based on pixel count and histogram-based segmentation is used for optic disc detection. Center of macula is detected based on estimated optic disc radius calculations and anatomical priors. In order to reduce computational burden, two disc diameter (DD) region near macula is extracted. Hard exudates are extracted with unique combination of CLAHE and morphological operations, followed by Kirsch's mask for sharp edge detection of hard exudates, and further morphological operations for noise removal. Results OD detection algorithm is evaluated on four different databases DiaretDB1, MESSIDOR, DRIONS-DB, and images from local hospital, overall sensitivity of 95.71% is achieved, highest sensitivity value is 98%, and is achieved on first 100 images from MESSIDOR database. Further DiaretDB1 and images from local hospital are processed further for macula detection and exudates finding in macular proximity, as DiaretDB1 is rich in variety of DR lesions and images from the local hospital to deal with real-time issues. Finding macula region is achieved with an overall success rate of 91.93% and overall sensitivity of 99% is reported on both the chosen databases. Conclusion Proposed work provides an automated tool, for early diagnosis of diabetic macular edema. The method is straightforward, robust and computationally less complex and improved success rates for feature, and lesion detections are achieved as compared with state-of-the-art methods.

show abstract

“…Hard exudates are bright lesions with well-defined edges but variable shapes. We adopt a method based on background estimation to detect hard exudates [ 19 ]. The bright objects including OD and hard exudates are acquired using background estimation, which is based on morphological reconstruction.…”

Section: Proposed Npdr Screening Systemmentioning

confidence: 99%

Automatic non-proliferative diabetic retinopathy screening system based on color fundus image

et al. 2017

Self Cite

View full text Add to dashboard Cite

BackgroundNon-proliferative diabetic retinopathy is the early stage of diabetic retinopathy. Automatic detection of non-proliferative diabetic retinopathy is significant for clinical diagnosis, early screening and course progression of patients.MethodsThis paper introduces the design and implementation of an automatic system for screening non-proliferative diabetic retinopathy based on color fundus images. Firstly, the fundus structures, including blood vessels, optic disc and macula, are extracted and located, respectively. In particular, a new optic disc localization method using parabolic fitting is proposed based on the physiological structure characteristics of optic disc and blood vessels. Then, early lesions, such as microaneurysms, hemorrhages and hard exudates, are detected based on their respective characteristics. An equivalent optical model simulating human eyes is designed based on the anatomical structure of retina. Main structures and early lesions are reconstructed in the 3D space for better visualization. Finally, the severity of each image is evaluated based on the international criteria of diabetic retinopathy.ResultsThe system has been tested on public databases and images from hospitals. Experimental results demonstrate that the proposed system achieves high accuracy for main structures and early lesions detection. The results of severity classification for non-proliferative diabetic retinopathy are also accurate and suitable.ConclusionsOur system can assist ophthalmologists for clinical diagnosis, automatic screening and course progression of patients.

show abstract

Hard Exudates Detection Method Based on Background-Estimation

Cited by 7 publications

References 13 publications

Exudates Detection Using Morphology Mean Shift Algorithm in Retinal Images

Exudates Detection Using Morphology Mean Shift Algorithm in Retinal Images

Automated macula proximity diagnosis for early finding of diabetic macular edema

Automatic non-proliferative diabetic retinopathy screening system based on color fundus image

Contact Info

Product

Resources

About