Automatic measurement of sister chromatid exchange frequency.

Abstract

An automatic system for detecting and counting sister chromatid exchanges in human chromosomes has been developed. Metaphase chromosomes from lymphocytes which had incorporated 5-bromodeoxyuridine for two replication cycles were treated with the dye 33258 Hoechst and photodegraded so that the sister chromatids exhibited differential Giemsa staining. A computer-controlled television-microscope system was used to acquire digitized metaphase spread images by direct scanning of microscope slides. Individual objects in the images were identified by a thresholding procedure. The probability that each object was a single, separate chromosome was estimated from size and shape measurements. An analysis of the spatial relationships of the dark-chromatid regions of each object yielded a set of possible exchange locations and estimated probabilities that such locations corresponded to sister chromatid exchanges. A normalized estimate of the sister chromatid exchange frequency was obtained by summing the joint probabilities that a location contained an exchange within a single, separate chromosome over the set of chromosomes from one or more cells and dividing by the expected value of the total chromosome area analyzed. Comparison with manual scoring of exchanges showed satisfactory agreement up to levels of approximately 30 sister chromatid exchanges/cell, or slightly more than twice control levels. The processing time for this automated sister chromatid exchange detection system was comparable to that of manual scoring.

Complex genetic control of host susceptibility to coxsackievirus B3-induced myocarditis

“…15,22 Briefly, mice were injected with EBD, which labels the damaged cells as visibly blue as well as making them fluoresce orange. Here we have adapted a program 35 to quantify the phenotype of percent EBD-positive surface area, termed sarcolemmal disruption. For each image, the program establishes a background brightness threshold and calculates the area of a tissue stained above this background as a percentage of total area ( Figure 3).…”

“…This was achieved as follows: each fluorescent image was captured digitally in grayscale where each pixel has an intensity value between 0 (black) and 255 (white). Then, a background intensity threshold was calculated by triangle algorithm 35 for each image through the use of the DIPimage package (http:// www.ph.tn.tudelft.nl/DIPlib/). Briefly, the threshold function used constructs a histogram of the frequency of each pixel intensity, and then a line is drawn between the highest frequency pixel and the farthest (brightest) pixel intensity on the histogram.…”

“…Briefly, the threshold function used constructs a histogram of the frequency of each pixel intensity, and then a line is drawn between the highest frequency pixel and the farthest (brightest) pixel intensity on the histogram. 35 The maximum distance between this line and the histogram is then set as a threshold as it corresponds to the highest staining intensities within normal tissue. In highly resistant mice, the lack of bright pixels lead to an underestimation of threshold value and a corresponding overestimation of percent tissue disrupted as determined by manual tracing.…”

Automatic scoring of sister chromatid exchanges by image analysis in a dose response experiment

“…Zack et al [1977] and Shafer et al [1980Shafer et al [ , 1986 located potential SCEs where separate darkly stained regions found by thresholding are closely adjacent. This method tends to find a false SCE at the centromere, whether there is in fact an exchange or not.…”

“…The feasibility of automatic scoring of SCEs has been demonstrated by several authors [Zack et al, 1976[Zack et al, , 1977Shafer et al, 1980;He and Chai, 1990], but not so far in the context of a complete system applied to routinely prepared samples. Except that Zack et aJ.…”