Table 5: Morphological features used in the classification experiment.

No.FeatureDescriptionNEN1LumenEN2

1 AreaArea of selection in square pixels * * *
2 PerimeterThe length of the outside boundary of the selection * * *
3 WidthWidth of the smallest rectangle enclosing the selection * * *
4 HeightHeight of the smallest rectangle enclosing the selection * * *
5MajorAxis Major (primary) axis length of the best fitting ellipse * * *
6MinorAxis Minor (secondary) axis length of the best fitting ellipse * * *
7 Circularity4π × (Area/Perimeter2). A value of 1.0 indicates a perfect circle * * *
8 Feret’s diameterThe longest distance between any two points along the selection boundary * * *
9 AspectRatioMajorAxis/MinorAxis * * *
10 SkewnessThe third order moment about the mean * * *
11 Roundness4 × Area/(π × MajorAxis2). The measure of the sharpness of particle’s edge and corners * * *
12 SolidityArea/ConvexArea * * *
13 RMSAARoot-mean-squared atypia-amplitude *
14 TSAVTotal sum of atypia volatilities for PIPs *
15 AtypiaRatioThe ratio of atypia region *
16 #AtypiaRegionsThe number of atypia regions for identifying papillary *
17CytoplasmLengthThe cytoplasm length of epithelial nucleus [32] *
18CytoplasmLengthSDThe standard deviation of CytoplasmLength *

NEN1: nonepithelial nuclei; EN2: epithelial nuclei.