Comparison of segmentation algorithms for fluorescence microscopy images of cells

Abstract

The analysis of fluorescence microscopy of cells often requires the determination of cell edges. This is typically done using segmentation techniques that separate the cell objects in an image from the surrounding background. This study compares segmentation results from nine different segmentation techniques applied to two different cell lines and five different sets of imaging conditions. Significant variability in the results of segmentation was observed that was due solely to differences in imaging conditions or applications of different algorithms. We quantified and compared the results with a novel bivariate similarity index metric that evaluates the degree of underestimating or overestimating a cell object. The results show that commonly used threshold-based segmentation techniques are less accurate than k-means clustering with multiple clusters. Segmentation accuracy varies with imaging conditions that determine the sharpness of cell edges and with geometric features of a cell. Based on this observation, we propose a method that quantifies cell edge character to provide an estimate of how accurately an algorithm will perform. The results of this study will assist the development of criteria for evaluating interlaboratory comparability. Published 2011 Wiley-Liss, Inc.y Key terms fluorescence microscopy; k-means cluster; image segmentation; cell edge; bivariate similarity index NUMEROUS areas of biomedical research rely on imaging of cells to provide information about molecular and phenotypic responses of cells to pharmaceuticals, toxins, and other environmental factors (1,2). Cell imaging is widely used in biological experiments because it can provide information on several relevant scales simultaneously. Molecular and supramolecular scales can be probed with the use of antibody and other specific affinity reagents and with fluorescent proteins. Gross phenotypic characteristics of cells that characterize their ultimate functional state can be examined in the same experiment, and often the temporal regime can be probed simultaneously. Together, these applications of cell imaging allow inference about the molecular details and the complex outcomes of the cellular biochemistry.Because of the enormous number of parameters that may influence a biological outcome, cell imaging experiments are often done in a ''high content '' mode (3,4), where large numbers of paired and replicate experiments are carried out simultaneously and result in very large (often gigabyte) image datasets. Such a large volume of image data precludes visual inspection of every image, and automated image processing and analysis is the only viable approach to data analysis.Segmentation of cell objects is a common image analysis operation that provides spatial and other features of identified objects and often precedes other operations to quantify parameters such as intracellular fluorescence. Segmentation can pose significant challenges to automated image processing and analysis. Because morphological features are often important in...

A Guided Tour of Selected Image Processing and Analysis Methods for Fluorescence and Electron Microscopy

“…As expansion ceases, contractile cortex reassembles under the membrane and drives bleb retraction. This is the case of amoeboid cells where membrane deformations at high frequency remind those from other cells where the plasma membrane detaches from the cytoskeleton, e.g., apoptotic cell [267], non-adherent carcinosarcoma cells [268], or melanoma cells showing prolonged blebbing while spreading [265]. These blebbing mechanisms are different of those described in [269], where no cortical breakdown prior to membrane bulging has been observed.…”

A Human Inspired Local Ratio-Based Algorithm for Edge Detection in Fluorescent Cell Images

“…Human manual outlines under study segmentation. Five-means clustering consistently under-estimates the cell area, as described in a paper currently in process [8]. To ensure that we include a broad band of pixels around the cell edge, we use an upper cutoff value that is 1 standard deviation above the 5-means clustering lowest centroid value, assuming that the lowest centroid represents the image background.…”

Severity Analysis of Cervical Cancer in Pap Smear Images by using EEETCM, ERSTCM & CFE method based Texture Features and Hybrid Kernel based Support Vector Machine Classifier.

“…Among them one of the most important method is the segmentation of cell nuclei from the stained specimens [14]. However, isolated nuclei of the cells in high-quality acquisitions provide difficult tricks in the segmentation is more number of nuclei with various characteristics under occurring different acquisition conditions in good-resolution scans of the complete microscope slide [15]. Thresholding is the most important method for segmentation process [16] and also it is an easiest method to convert gray scale image into binary image based on the global or local threshold value [17,[18].…”