Premium
Ovarian dysplasia: Nuclear texture analysis
Author(s) -
Deligdisch Liane,
Miranda Carlos,
Barba Joseph,
Gil Joan
Publication year - 1993
Publication title -
cancer
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 3.052
H-Index - 304
eISSN - 1097-0142
pISSN - 0008-543X
DOI - 10.1002/1097-0142(19931201)72:11<3253::aid-cncr2820721120>3.0.co;2-1
Subject(s) - dysplasia , medicine , pathology , malignancy , texture (cosmology) , nuclear medicine , artificial intelligence , computer science , image (mathematics)
Background. Ovarian dysplasia has been defined by histologic 1,2 and morphometric studies 3,4 focusing on architectural and nuclear profile changes. A new technique is used to enhance the accuracy of this diagnosis by a quantitative evaluation of the nuclear texture that represents the nuclear chromatin pattern on which conventional diagnoses of malignancy are usually made. Methods. Histologic sections from 35 ovaries classified as malignant (12), dysplastic (12), and normal (11) were evaluated by tracing boundaries of nuclear profiles and measuring “textons” (texture primitives) with a histogram analysis of three zones of gray level densities (called for simplification white, gray, and dark). The average combined area was tabulated for specimens with the same diagnosis, and linear regression plots compared the texton area with total nuclear area. Results. The dimensions of textons originally hidden inside the chromatin and revealed by histograms were found to be closely clustered in normal epithelium, and increasingly dissociated from the containing nucleus as the lesion progressed from dysplastic to malignant. The statistical multivariate analysis including nine parameters correctly classified the three diagnostic categories as normal, dysplastic, and malignant. Conclusions. Computerized image analysis of nuclear texture adds accuracy to the recently elaborated morphometric methods to define ovarian dysplasia, a potential precursor of ovarian carcinoma.