Skip to main content
. 2018 Mar 9;8:4311. doi: 10.1038/s41598-018-22489-1

Figure 5.

Figure 5

Intensify3D restores distorted artificial 3D data and facilitates quantification (a). 3D rendering of undistorted artificial 3D image stack of ~500 Gaussian spheres. Image stack dimensions – X Y Z: 600 × 600 × 500 pixels (b). Representative image (Z = 50) of undistorted image stack (left bottom) and distorted image stacks before and after correction with Intensify3D or CIDRE. All images are presented at the same minimum/maximum brightness levels. Outline of intensity histogram for the undistorted image (red curve) is overlaid on top of black filled intensity histograms for each of the individual images (left bottom corner) (c). ROC space for true/false positive detection rates by 3D object counter plugin (FIJI) on distorted data (crosses), Intensify3D (filled circles) or CIDRE (empty circles) corrected. Performance is in comparison to undistorted data (d). Relative mean absolute error for measured statistics of true positive spheres for distorted data, Intensify3D or CIDRE corrected. Performance is in comparison to undistorted data. Each row was corrected so that the minimal error is 1.