Skip to main content
Figure 4 | BMC Bioinformatics

Figure 4

From: Dual channel rank-based intensity weighting for quantitative co-localization of microscopy images

Figure 4

Effect of intensity correlations on various co-localization algorithms. A synthetic image composed of a 16 segment sub-grid, each of different intensity, was generated (channel A). The same image was duplicated and rotated at various angles (channel B). These images were analyzed by Manders' (M), Costes' (MC) and RWC algorithms. In the merged images, the Costes' mask (considered for co-localization analysis) is shown in white with a blue border. The remaining regions of the images are shown in their corresponding LUTs. The co-localization coefficients, and thresholds used (in parentheses) are also given. For calculation of the Manders' and RWC coefficients, the threshold was set at either 15% of the maximum intensity in each channel (black text), or no thresholds were applied (cyan text). (A) Two channels containing identical data result in co-localization coefficients of 1.0 in all cases. (B) The Costes' mask disregarded 62.5% of the image, despite the presence of high pixel intensities in these areas which should contribute to the overall quantification. (C) The Costes' mask disregarded the entire image because the intensities across the image were anti-correlating between the two channels. (D) The Costes' mask disregarded 87.5% of the image due to the varying correlation of intensities between the two channels. In all examples the RWC coefficients were as expected, based on the intensity distribution of the synthetic images. When no thresholding was applied the Manders' algorithm reported co-localizations coefficients of 1.0, whereas the RWC coefficients remained similar to those observed when thresholding.

Back to article page