To simplify the prototyping process, we could generate two images, one per channel, containing simple shapes (circles, rectangles etc…) which sizes, and degree of overlap are controlled. In microscopy, the images result from single captions of a scene, which are impaired by the optics and corrupted by noise. This somewhat barbarian terms designate a computer-generated dataset where all parameters are human controlled. When building a workflow, the image analyst might benefit from use of a synthetic dataset. In addition to the requirement of a proper quantification strategy, this example also points out the need for a well characterized dataset. 2013).ģ.1.3 Some Image Preprocessing Tips You Should Keep in Mind
Recently a new analytical statistics method based on Ripley’s K function is proposed and included as an Icy plugin, Colocalisation Studio (Lagache et al. ( 2010) but it is computationally expensive.
A common method is Monte-Carlo simulations Fletcher et al. 3.4 and 3.5.Ĭolocalisation studies generally should perform some statistical analysis, in order to interpret whether the found co-occurrence or correlation is just a random coincidence or a true colocalisation. We will implement some specific methods for both categories in two case studies described in ► Sects. 2013), or if two objects with certain percentage of area/volume overlap (Rizk et al. Usually, two objects are considered colocalised, if the centroids of the objects are within certain distance (Bolte and Cordelières 2006 Cordelières and Bolte 2014 Obara et al. Most object-based colocalisation methods first segment and identify objects, and then account for objects’ inter-distances to analyze possible colocalisation.
Tools for quantifying these measures can be found in many image analysis open-source and commercial software packages, to name just a few: Fiji’s JACoP plugin and Coloc 2, CellProfiler, BioImageXD, Huygens, Imaris, Metamorph, Volocity. Several review papers have been published during the last decade, where coefficients’ meaning, interpretation, guide of use, and examples for colocalisation are given (Bolte and Cordelières 2006 Cordelières and Bolte 2014 Dunn et al. Intensity based methods compute global measures about colocalisation, using the correlation information of intensities of two channels. In general, two categories of quantitative approaches to colocalisation analysis can be found: intensity based correlation methods and object based methods. But visual evaluation does not tell the degree of colocalisation, nor if it is true colocalisation at all. the points form a straight line if the two structures correlate. Scatter plot of pixel intensities from the two images also qualitatively indicates colocalisation, e.g. However, this intuitive method can work only when the intensity levels of the two images are similar (see a detailed example in Dunn et al. For example, colocalisation of red and green structures can appear yellow. It may be identified by superimposing two images and inspecting the appearance of the combined color.