NUCLEAR PATTERN-RECOGNITION BY 2-PARAMETER TEXTURE ANALYSIS

The present paper describes a simple procedure for the analysis of chr omatin texture. High-resolution digitized images of nuclei are first s tandardized to render gray values invariant to staining and illuminati on conditions. Subsequently, the nucleus is subdivided by a square gri d into 0.4 x 0.4 mu m(2) quadrats and standard deviations of gray valu es within each quadrat are estimated. Finally, the overall mean and st andard deviation of quadrat standard deviations are calculated. These values may be considered as pure descriptors of the nuclear texture, a s they represent the distribution of chromatin changes, disregarding a ny absolute densitometric and morphometric feature. Using the above de scriptors it is possible to recognize at least seven chromatin pattern s in a mixed population of developing and degenerating neurons. Result s are visually verified by mapping the original pictures at the corres ponding bivariate plot points. Comparison with the Markovian texture a nalysis is discussed.