Coefficient of variation as an image-intensity metric for cytoskeleton bundling

Abstract

The evaluation of cytoskeletal bundling is a fundamental experimental method in the field of cell biology. Although the skewness of the pixel intensity distribution derived from fluorescently-labeled cytoskeletons has been widely used as a metric to evaluate the degree of bundling in digital microscopy images, its versatility has not been fully validated. Here, we applied the coefficient of variation (CV) of intensity values as an alternative metric, and compared its performance with skewness. In synthetic images representing extremely bundled conditions, the CV successfully detected degrees of bundling that could not be distinguished by skewness. On actual microscopy images, CV was better than skewness, especially on variable-angle epifluorescence microscopic images or stimulated emission depletion and confocal microscopy images of very small areas of around 1 μm2. When blur or noise was added to synthetic images, CV was found to be robust to blur but deleteriously affected by noise, whereas skewness was robust to noise but deleteriously affected by blur. For confocal images, CV and skewness showed similar sensitivity to noise, possibly because optical blurring is often present in microscopy images. Therefore, in practical use with actual microscopy images, CV may be more appropriate than skewness, unless the image is extremely noisy.

Implicit Laplacian of Enhanced Edge: An Unguided Algorithm for Accurate and Automated Quantitative Analysis of Cytoskeletal Images

“…For bundling indices, we were interested in their level of conformity because direct indices (based on binary shape) and indirect indices (based on brightness frequency distribution) are entirely different strategies to measure bundling. Using the same approach of correlation analysis, we found that diameter_TDT and diameter_SDT display strong positive correlation, while skewness and CV have merely medium-low correlation, which echoes the previous report demonstrating skewness and CV have different performance on the bundling evaluation 20 .…”

“…Through the image processing pipeline, cytoskeletal indices were automatically calculated from the binary image generated by ILEE. As a substantial expansion from three previously defined cytoskeletal indices (e.g., occupancy, skewness , and CV ) 17,20 , we totally introduced 12 indices (Fig. 1a); particularly, we focused on 9 of the 12 indices in this study, as they are normalized and ready for biological interpretation (see Supplementary Fig.…”

“…While this approach utilizes manual global thresholding (MGT), which can potentially introduce a certain level of user bias, it still outperforms most standardized adaptive/automatic global or local thresholding approaches, such as Otsu 18 and Niblack 19 . More recent advances in MGT-based approaches, such as those by Higaki and colleagues, include additional analysis indices, such as the introduction of coefficient of variation ( CV ) of fluorescence to quantify the degree of filament bundling – an improvement that advances the overall robustness and utility of the original algorithm 20 . However, this analysis pipeline still consumes a considerable amount of time and effort from users for massive sample processing and leaves unaddressed two critical issues of rigor in image processing and analysis: information loss and human bias.…”

Filament organization of the bacterial actin MreB is dependent on the nucleotide state

“…4 B and Video 1 ). Quantification of the spatial organization, by measuring anisotropy using FibrilTool ( Boudaoud et al, 2014 ), and coefficient of variation (CV; Higaki et al, 2020 ), which is an indicator of cytoskeleton bundling, further confirmed that ScMreB5 E134A exhibited differences in bundling of filaments ( Fig. 4, C and D ).…”

Estimating and evaluating the rice cluster distribution uniformity with UAV-based images

“…To eliminate the influence of the rice cluster density, this paper adopted the coefficient of variation (CV) as a heterogeneity index for evaluating the heterogeneity in the rice cluster distribution in the region of interest 29 . where is the area of the region occupied by the rice clusters, is the number of rice clusters, is the mean.…”