An automated cell viability quantification method for low-resolution confocal images of closely packed cells based on a modified gradient flow tracking algorithm

作者：Kaviani R; Merat P; Moldovan F; Villemure I^*

来源：Journal of Microscopy, 2016, 261(3): 217-226.

DOI：10.1111/jmi.12322

• Convergence centers
• gradient flow tracking
• live
• dead labelling
• viability quantification

Fluorescent-based live/dead labelling combined with fluorescent microscopy is one of the widely used and reliable methods for assessment of cell viability. This method is, however, not quantitative. Many image-processing methods have been proposed for cell quantification in an image. Among all these methods, several of them are capable of quantifying the number of cells in high-resolution images with closely packed cells. However, no method has addressed the quantification of the number of cells in low-resolution images containing closely packed cells with variable sizes. This paper presents a novel method for automatic quantification of live/dead cells in 2D fluorescent low-resolution images containing closely packed cells with variable sizes using a mean shift-based gradient flow tracking. Accuracy and performance of the method was tested on growth plate confocal images. Experimental results show that our algorithm has a better performance in comparison to other methods used in similar detection conditions.
Lay description In biology, it is frequently needed to count the number of live and dead cells in microscopic images. When processing large number of samples, automatic cell counting approaches become very useful. Most of the available viability tests results in fluorescent images. Therefore, many algorithms have been previously proposed for identifying and counting the number of cells in such images. Although several methods have been previously developed for cell segmentations in images with closely packed cells of various size and shapes, but most of them fail to identify cells in low-resolution images. The purpose of ours study was to develop and validate an algorithm based on gradient flow tracking, which is able to count the number of closely packed cells with different size and shapes in low resolution images. Our method is based on finding the centers of cells as an indicator of the number of cells in an image by detecting the intersection of gradient vectors. In comparison to other existing methods, our proposed algorithm resulted in a more accurate quantification of cells.

• 出版日期2016-3
• 单位McGill