Morphometry Features

 

Size Features

Nuclear Area:  area in square microns of the nucleus

Pg. DNA:  the DNA Mass in picograms of the nucleus

Aborbance (Density):  computed by dividing the nuclear mass (pg. DNA) by its area (pg.DNA / Area)

Avearge Absorbance (Average Optical Density):  summed optical density divided by the number of pixels comprising each nucleus (Sum OD / # pixels)

Summed Absorbance (Summed Optical Density):  sum of each individual optical density value (grayscale value) over all the pixels comprising the nucleus

 

Shape Features

Perimeter:  perimeter of the nuclear border in microns

Elongation:  computed by dividing the maximum diameter by the minimum diameter (Max Diam / Min Diam)

Maximum Diameter:  maximum diameter of the nucleus in microns
Minimum Diameter:  minimum diameter of the nucleus in microns

Cell Ferret X:  width of a bounding rectangular box around the nucleus object, measured in microns
Cell Ferret Y:  Height of a bounding rectangular box around the nucleus object, measured in microns
Shape:  computed as the perimeter squared divided by the area (Perimeter / Area)

 

General Texture Features

Run Length:  count of pixels within the nucleus whose gray level values differ form those of their left and right neighbors

Configurable Run Length:  count of pixels within the nucleus whose gray level values differ from those of their left and right neighbors (Pixel Sample size and Pixel Difference Threshold set up by the operator)

Valley Slope Peak
Valley, Slope, Peak example

Valley: average number of valleys appearing on the surface of the nucleus. A valley is defined as both  neighboring pixels having higher gray level values than the current pixel.

Slope: defined as one neighbor pixel having a lower gray level value and the other neighbor pixel having a higher gray level value than the current pixel.

Peak:  average number of peaks appearing on the surface of the nucleus. A peak is defined as both neighboring pixels having lower gray level values than the current pixel.

Standard Deviation:  the standard deviation of optical density of every pixel that composes the nucleus

 

Markovian Texture Features

Markovian texture parameters are a quantitative characterization of gray level variations between adjacent pixels in an image.  Parameters are constructed in 4 steps:

    • The Optical Density histogram is normalized into 8 equally frequent density levels.
    • A gray level transition matrix is computed by comparing every pixel with its neighboring pixel.
    • The transition matrix is transformed into a relative frequency (probability) matrix, M, where each entry,  PL(i|j), is the probability that gray level i will occur L steps after gray level j.
    • Various texture parameters are computed from matrix M. 

Angular Second Moment
Contrast
Correlation
Difference Moment
Inverse Difference Moment
Sum Average
Sum Variance
Sum Entropy
Entropy
Difference Variance
Difference Entropy
Information Measure A
Information Measure B
Maximal Correlation Coefficient
Coefficient of Variation
Peak Transition Probability
Diagonal Variance
Diagonal Moment
Second Diagonal Moment
Product Moment
Triangular Symmetry

 

 

 

Colorado State University CHOICE BreastWatch