Discriminant Analysis of Tumor Cell Subpopulation Based on Morphometric and Photometric Features: Observations on tumor cells of the uterine cervix carcinoma.
- Author:
Chang Soo PARK
;
Dae Yong CHOI
;
Min Cheol LEE
;
Sang Woo JUHNG
- Publication Type:Original Article
- Keywords:
Discriminant analysis;
DNA ploidy;
Morphometry;
Photometry
- From:Korean Journal of Pathology
1993;27(2):108-114
- CountryRepublic of Korea
- Language:Korean
-
Abstract:
DNA aneuploid cells are poorly characterized in both biochemical and morphological terms. This study was performed to see the relationship between DNA ploidy and morphometric and photometric nuclear features. DNA contents of tumor cells were measured by image cytometry in 46 cases of micro- or early invasive squamous cell carcinoma of the uterine cervix. Also measured were nuclear area, perimeter, maximum diameter, chromatin pattern index, and staining intensity. Among the 46 cases, 20 cases which had both DNA diploid and aneuploid cell subpopulations were selected, and the two subpopulations were discriminated statistically. Multivariate discriminant analysis seperated clearly the two subpopulations, whereas univariate analysis failed. For canonical discriminant function, nuclear area was selected first, followed by staining intensity in each case. Other variables selected afterwards were nuclear perimeter, maximum diameter, and/or chromatin pattern index in random fashion. Correlation coefficient between the canoncial discriminant function and the variables were 0.20~0.40 for nuclear area and 0.25 or less for the others. The above results suggest that DNA ploidy is a parameter more or less independent on individual morphometric and photometric parameters.
