Table 8. The techniques and datasets used for RBC classification.

Author	Database	Segmentation	Features	Enhancement	Classification
Du et al. (2019)	17,933 samples from the hospital	ImageNet model	Morphological feature	Nil	CNN model
Sampathila, Shet & Basu (2018)	Leishman-stained thin blood smear slides	RGB color based	GLCM	Color space	GUI
Kihm et al. (2018)	4,000 Manually classified images	Nil	Shape, size	By convolution of NN	CNN
Imran & Ahmad (2017)	ALL-IDB1 108 images	Statistical based thresholding	Morphological	Rayleigh distribution	SVM and ELM
Das, Maiti & Chakraborty (2018)	950 blood cells	Marker-controlled watershed algorithm	Mean intensity, standard deviation, skewness, kurtosis and entropy	Special Fuzzy C-mean	Random forest
Yi, Moon & Javidi (2016)	117 images Manual collection	Marker-controlled watershed transform	Area, perimeter, circulatory etc.	Watershed transform algorithm	Gabor-filtered holographic
Abood, Karam & Hluot (2017)	Nil	Clustering	Shape and color		Fuzzy logic
Xu et al. (2017)	7,000 single RBCs	Nil	Shape	Geometric transformations	Deep CNN
Acharya & Kumar (2017)	1,000 images manually collected	K-medoids algorithm	Area, perimeter, diameter, shape, geometric	Nil	Modified watershed transform