Cell migration plays a major role in development, physiology, and disease, and is frequently evaluated in vitro by the monolayer wound healing assay. The assay analysis, however, is a time-consuming task that is often performed manually. In order to accelerate this analysis, we have developed TScratch, a new, freely available image analysis technique and associated software tool that uses the fast discrete curvelet transform to automate the measurement of the area occupied by cells in the images. This tool helps to significantly reduce the time needed for analysis and enables objective and reproducible quantification of assays. The software also offers a graphical user interface which allows easy inspection of analysis results and, if desired, manual modification of analysis parameters. The automated analysis was validated by comparing its results with manual-analysis results for a range of different cell lines. The comparisons demonstrate a close agreement for the vast majority of images that were examined and indicate that the present computational tool can reproduce statistically significant results in experiments with well-known cell migration inhibitors and enhancers.

Introduction

Cell migration has a major role in development and in physiological repair processes, and is also involved in many pathological disorders, including cancer invasion and metastasis, angiogenesis, and inflammatory reactions (1, 2).

Assessment of the migrative potential of distinct cell types is important for the basic understanding of the molecular mechanisms involved, but also for screening of pharmaceutical compounds that modulate cell migration and thereby might exert beneficial effects in pathological conditions (3, 4). Several assays have been developed to measure and compare cell motility in vitro, among them the Boyden chamber transmigration assay (5). The most widely used technique is the in vitro monolayer wound assay, or the “scratch assay” (5). In this assay, a confluent cell layer grown on a plastic support is artificially wounded and the coverage of open area by migrating cells is then assessed over time using bright-field imaging (6). The limitation of this method is the manual and highly subjective nature of open-area quantification (7). Overcoming this limitation will allow users to significantly accelerate analysis and translate this widely used manual assay to an automated, quantitative high-throughput system (8).

An important requirement for reliable and reproducible quantification of the open area is a robust and flexible image segmentation algorithm. So far, the performance of available algorithms has been limited by the strongly varying image quality, which depends on the acquisition mode and skills of the experimenter. Better-performing algorithms are not freely available. In this work, we applied a novel multi-purpose image analysis method based on curvelets (9, 10) in combination with an automated thresholding scheme to overcome these limitations. This combination allows customization of the quantification process and provides a consistent and efficient image analysis framework.

Here, we present TScratch, a user-friendly software tool that implements our analysis algorithm for monolayer wound healing assays, and which is freely available (www.cse-lab.ethz.ch/software.html). The software enables a reliable and reproducible quantification of open areas and an immediate readout for a broad spectrum of cell lines.

Materials and methods

Image segmentation and thresholding

We developed a novel algorithm for measuring the open image area based on a method which, for the first time in biomedical image analysis, uses the recently developed edge-detection algorithm (10) based on the discrete curvelet transform (9). The curvelet transform encodes image information for different scales, directions and positions in the form of curvelet coefficients. The transform is lossless and when all curvelet coefficients are maintained, the original image can be reconstructed. The curvelet coefficients can then in turn be processed so that the desired image properties are obtained by using different scales and orientations. For the presented application, 2 scale levels—corresponding to the levels of detail at which the cells’ boundaries appear in the image—are selected. At each position, the magnitudes of the curvelet coefficients at the different scales and directions are added together. This generates a curvelet magnitude image, with the intensities providing a measure for the amount of detail in the original image, thereby helping to separate vacant and occupied regions depending on the level of detail. In order to further enhance the separation and minimize the sensitivity to local variations, the algorithm then computes the morphological opening (an erosion followed by a dilation) of this curvelet magnitude image, using a disk of adjustable radius as the structural element of the opening operation.

Following this processing, a threshold is applied to separate the 2 regions by setting the threshold in a gap between 2 peaks in the histogram of the curvelet magnitude image, thus separating the low-intensity open area and the high-intensity covered area. In cases of very small, open areas, it is harder to determine the threshold automatically, and a default threshold is applied, which the user can modify if desired.

Additionally, since the quantification of contiguous open and occupied areas is desired, small, isolated open areas are marked as occupied and small isolated occupied areas are marked as open, based on a threshold parameter in the settings menu that may be adjusted by the user. An erosion step is also performed to make the edges of the marked occupied area adapt better to the cell boundaries.