Skip to main content

Table 1 Four categories of features were extracted and tested in the automatic pipeline

From: Ranked retrieval of segmented nuclei for objective assessment of cancer gene repositioning

Category

Name

Brief description

 

Area

Number of pixels in segmentation mask

 

Perimeter

Length of segmentation boundary

 

Perimeter/Area

Perimeter to area ratio

 

Solidity

Ratio of area to the convex hull area

 

CH Perimeter / Perimeter

Ratio of convex hull perimeter to object perimeter

 

Max Circle Ratio

Ratio of the area of the largest inscribed circle to the total area

 

Ellipse Eccentricity

Eccentricity of the ellipse that has the same second-moments as the mask

Morphological

Ellipse Error Ratio

Area difference between best-fit ellipse and boundary

Length

Measure of elongation

 

Width

Measure of breadth

 

Mean Pairwise Distance

Mean all-pairs distance between points on perimeter

 

Polar Histogram

Measures isotropy of border points at a given angle

 

Num. Severe Corners

Number of strong corners in segmentation contour

 

Box-counting Dimension

Fractal dimension of the perimeter

 

Erosion Profile

Identifies segmentations with narrow passages separating large areas

 

Elliptical Fourier

Number of elliptical Fourier coefficients to reconstruct the mask to within

  

10% area error

 

Mean Intensity

Mean of grayscale intensity inside nucleus

Texture

Intensity Range

Range of grayscale intensity inside nucleus

Entropy

Global entropy of grayscale values inside nucleus

 

Gray-level Co-occurrence

Statistics of the gray-level co-occurrence matrix

 

Num. FISH

Number of FISH spots

 

FISH/Area

Number of FISH spots normalized by area

FISH

FISH CH Area

Ratio of convex hull area formed on FISH spots to total area

 

FISH Boundary

Measures whether the FISH convex hull intersects the nuclear boundary

 

Mean FISH Distance

Mean distance between FISH spots

 

Num. Nuclei

Number of candidate nuclei in the image

 

Intensity Ratio

Mean intensity of band surrounding the nucleus compared to the mean intensity inside

Contextual

Betweenness Centrality

Betweenness centrality of nucleus in a graph connecting nuclei in the image

Number Neighbors

Number of neighbors connected to the nucleus

 

Mean Edge Distance

Mean edge distance to 1st-level neighbors

  1. aSummarized here are brief descriptions of the features.