A Review on Recent Developments for Detection of Diabetic Retinopathy

Table 6

Different methods for detection of microaneurysms.

Algorithm	Image processing techniques	Database	Color space	Sensitivity	Specificity	Accuracy
Sopharak et al. [53]	Median filter, contrast enhancement, Shade Correction, and extended minima transform	Patient data	Green channel	81.61%	99.9%	99.98%
Krishna et al. [54]	Ensemble-based microaneurysms; detector, Walter Klein, and CLACHE	Messidor	Gray scale	—	—	—
Roy et al. [55]	Canny edge detection, morphological; reconstruction	DIARETDB1	Green channel	89.5%	82.1%	—
Adal et al. [56]	Contrast enhancement technique, Hessian-based candidate selection algorithm, and SVM classifier	ROC	Green channel	—	44.64%	—
Datta et al. [57]	Contrast Limited Adaptive Histogram; Equalization, median filter, and Image Catenation	Private data	Green channel	—	82.64%	99.98%
Giancardo et al. [58]	Radon cliff operator	ROC	Gray scale	41%	—	—
Ding and Ma [59]	Dynamic multiparameter template (DMPT) matching scheme	ROC		96%	—	—

ROC: retinopathy online challenge.