Multiparametric discrimination of serous ovarian tumors by analytical morphometry.

In order to enhance the discrimination power in the field of serous ovarian tumors, we applied the software system SAM (Shape Analytical Morphometry) to the analytic studies of biological forms. Besides the usual dimensional evaluations (perimeter, area, maximum diameter and shape index), this procedure permits the description of the nuclear form using analytical parameters: 1) extraction of nucleus fundamental curve; that is a functional curve giving the "smoothing" of the original contour by two parametric equations (separately for x and y values as independent variables); 2) evaluation of nuclei contour irregularities by Fourier analysis; 3) evaluation of shape asymmetry by SAE (Shape Asymmetry Evaluator); that is the ratio between the length of a segment of a parabola interpolating the original curve points, and a straight line joining its extremities for a 180 degrees barycentric rotation according 10 degrees steps. All parameters resulted to be independent and were submitted to multivariate discriminant analysis. We studied 180 nuclei from 18 cases of serous ovarian tumors, (6 benign, 6 borderline and 6 malignant tumors). With respect to the dimensional parameters, the application of analytical morphometry permitted us to reduce the minimum percentage error in the discrimination of the different classes. In fact, in the distinctions of benign and malignant nuclei, the minimum percentage error was 13.30%, against the 18.3% error when using dimensional morphometry. Furthermore, in the comparison of malignant and borderline nuclei there was a reduction of error from 23.3% to 22.5%, and in the comparison of benign and borderline nuclei, the error was reduced from 37.5% to 30%.(ABSTRACT TRUNCATED AT 250 WORDS)