Convolutional Neural Networks for Recognition of Lymphoblast Cell Images

Table 3

Summary of features extracted using image processing.

No.	Feature	ROI	Type	Description
1	N/C ratio	Cellular	Geometric	Ratio of number of pixels in nucleus to those in cytoplasm
2	Form factor	Nucleus		Ratio of number of pixels in nucleus to its perimeter
3	Roundness	Nucleus		Measurement of how nucleus shape is close to a circle
4	Eccentricity	Nucleus		Ratio of major axis to minor axis
5	Compactness	Nucleus		Degree to which a shape is compact
6	Symmetry	Nucleus		Ratio between two parts around the nucleus major axis
7	Hand-mirror	Cellular		Measurement of how the hand-mirror part of the cell forms
8	Fractal geometry	Nucleus		Degree to which the nucleus boundary is irregular by calculating Hausdorff dimension
9–11	Contour	Nucleus		Variance, skewness, and kurtosis of distances between centroid and contour points along the nucleus boundary
12	Fractal geometry	Cellular		Degree to which the cellular boundary is irregular by calculating Hausdorff dimension
13–15	Contour	Cellular		Variance, skewness, and kurtosis of distances between centroid and contour points along the cellular boundary
16–18	Haar wavelet	Nucleus	Texture	Mean of , and
19–21	Haar wavelet	Nucleus		Variance of , and
22–26	Haralick	Nucleus		Contrast, correlation, homogeneity, energy, and entropy of Haralick’s texture feature values
27–34	Fourier descriptors	Nucleus		Mean, standard deviation, skewness, and kurtosis of the frequency components obtained from discrete forward (27–30) and inverse (31–34) Fourier transforms
35–37	Color in RGB	Nucleus	Color	Mean color intensity of red, green, and blue in a nucleus area
38–40	Color in HSV	Nucleus		Mean color intensity of hue, saturation, and value in a nucleus area
41–43	Color in RGB	Cytoplasm		Mean color intensity of red, green, and blue in a cytoplasm area
44–46	Color in HSV	Cytoplasm		Mean color intensity of hue, saturation, and value in a cytoplasm area