G Büsche1, J Schlué, A Georgii. 1. Pathologisches Institut der Medizinischen Hochschule, Hannover, Germany.
Abstract
OBJECTIVE: A new parametric method is presented, called "square sampling," which speeds up the estimate of the number of cells or particles that are randomly distributed within a tissue. STUDY DESIGN: The principle of square sampling is subdivision of a biopsy into at least 100 squares of the same size using a measuring ocular or computer-based morphometric system and estimating the cell number by counting "positive" squares, squares with at least one cell of interest, assuming a binomial distribution of positive squares, depending on numerical density. RESULTS: The derived estimate yielded almost identical results when compared with the exact count of pseudo-Gaucher cells within bone marrow biopsies from untreated patients with chronic myeloid leukemia (r = .97, examined area = 94 x 2 mm2, with 400 squares/2 mm2), but (1) the total time of investigation could be halved by square sampling (25.1 versus 55.3 hours, P < .00005), and (2) the estimated number of cells did not very more widely around the mean exact count than the cell numbers exactly counted (P > .05). CONCLUSION: Square sampling is an easy, fast and effective alternative to nonparametric approaches in order to quantify the numerical density of cells randomly distributed within a tissue. The method can also be applied to test hypotheses of random distribution as well as to quantify a clustering of cells in cases of nonrandom cell distribution.
OBJECTIVE: A new parametric method is presented, called "square sampling," which speeds up the estimate of the number of cells or particles that are randomly distributed within a tissue. STUDY DESIGN: The principle of square sampling is subdivision of a biopsy into at least 100 squares of the same size using a measuring ocular or computer-based morphometric system and estimating the cell number by counting "positive" squares, squares with at least one cell of interest, assuming a binomial distribution of positive squares, depending on numerical density. RESULTS: The derived estimate yielded almost identical results when compared with the exact count of pseudo-Gaucher cells within bone marrow biopsies from untreated patients with chronic myeloid leukemia (r = .97, examined area = 94 x 2 mm2, with 400 squares/2 mm2), but (1) the total time of investigation could be halved by square sampling (25.1 versus 55.3 hours, P < .00005), and (2) the estimated number of cells did not very more widely around the mean exact count than the cell numbers exactly counted (P > .05). CONCLUSION: Square sampling is an easy, fast and effective alternative to nonparametric approaches in order to quantify the numerical density of cells randomly distributed within a tissue. The method can also be applied to test hypotheses of random distribution as well as to quantify a clustering of cells in cases of nonrandom cell distribution.
Authors: G Büsche; H Majewski; J Schlué; S Delventhal; S Baer-Henney; K F Vykoupil; A Georgii Journal: Virchows Arch Date: 1997-02 Impact factor: 4.064