Statistical analysis of molecule colocalization in bioimaging

Abstract

The quantitative analysis of molecule interactions in bioimaging is key for understanding the molecular orchestration of cellular processes and is generally achieved through the study of the spatial colocalization between the different populations of molecules. Colocalization methods are traditionally divided into pixel-based methods that measure global correlation coefficients from the overlap between pixel intensities in different color channels, and object-based methods that first segment molecule spots and then analyze their spatial distributions with second-order statistics. Here, we present a review of such colocalization methods and give a quantitative comparison of their relative merits in different types of biological applications and contexts. We show on synthetic and biological images that object-based methods are more robust statistically than pixel-based methods, and allow moreover to quantify accurately the number of colocalized molecules. V C 2015 International Society for Advancement of Cytometry

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

“…Moreover, related books [4], [5] and special issues [6]- [9] include a few tutorial-style overview articles covering progress in recent years for a large variety of topics (e.g., tracking in fluorescence bioimaging [10]- [12], sub-diffraction limited imaging and single molecule localization [13], [14], parametric active contour-based image segmentation [15] ). Finally, several authors presented independently state of the art methods for specific and important topics including cell-shape analysis [16], neuron tracing [17], co-localization (percentage of co-detection of interacting protein types at the same location) [18], [19], 3D image deconvolution [20], spot detection [21] in fluorescence microscopy Even if it is generally a difficult task to present a broad view of activities in bio-image processing and analysis [2], several authors [22]- [25] already explained successfully how computer vision, image analysis and visualization algorithms combined in workflows, will play a significant role in image-based studies of cell biology.…”

“…In the literature two categories of co-localization approaches are generally considered, which are either intensity-based or segmentation-based [18], [19]. The occurrence of yellow signals in an overlay fluorescence image generally depicts the correlation between the locations of the green and red signals in the cell, thus showing some co-localization between the two proteins under study.…”

“…Correlation measurements (Manders' descriptors [229]) are conventionally used in fluorescence microscopy. In order to quantify the above phenomenon, object-based methods have been applied in the literature (see [230] and [231], [19], where the spots detected as described above, are reduced to points and their interaction is analyzed by spatial statistics methods (point processes, descriptive Ripley's functions [232]). In this area, False Discovery Rate framework to test a set of distances between pairs of molecular entities could be more investigated [233].…”

Integrating cell biology, image analysis, and computational mechanical modeling to analyze the contributions of cellulose and xyloglucan to stomatal function

“…To reduce user bias and enhance quantitation, colocalization analyses can be automated through two basic approaches: pixel-based (signal overlap, intensity correlation), and object-based (spatial evaluation). 17 The latter is favored since it presents an opportunity for spatial exploration of the colocalized signal while being robust to noise and background fluorescence. 18 6 overall stress also increases but concentrated stress near the opening decreases, suggesting feedback between mechanical stress and wall deposition to reinforce the high stress area.…”

Advancing multiscale structural mapping of the brain through fluorescence imaging and analysis across length scales

“…Object-based measures of co-localization and nonlinear correlation metrics such as mutual information that have found widespread application in other fields may also prove valuable for extracting more subtle spatial relationships that are not captured by the linear Pearson and Manders' metrics. For example, Lagache et al [62] found that object-based Ripley's K-function for co-localization showed improved robustness to noise relative to Manders' and Pearson metrics when tested on dual-channel synthetic images with Gaussian noise. As new experimental techniques enable the recording of numerous molecularly specific channels, the adoption of multiple comparison procedures will be critical to examining many pairwise relationships in these multiplexed co-localization studies.…”