Image texture features for cell nuclei based on absorbance surface curvature.
The absorbance values of a cell nucleus, as measured at each pixel location, can be considered as forming a surface in a three-dimensional space. The principal curvatures of differential geometry can then be evaluated for this surface. From these curvatures, two new variables are computed that, when averaged over all pixels in the nucleus, are viewed as image texture measures. These two measures were found to be effective for differentiating insect populations and compared well with other features used in the TICAS system. This paper presents the results of comparing the performance of these two features with the performance of other features for several populations of cells.