FastSME: Faster and Smoother Manifold Extraction from 3D Stack

Abstract

3D image stacks are routinely acquired to capture data that lie on undulating 3D manifolds yet processed in 2D by biologists. Algorithms to reconstruct the specimen morphology into a 2D representation from the 3D image volume are employed in such scenarios. In this paper, we present FastSME, which offers several improvements on the baseline SME algorithm which enables accurate 2D representation of data on a manifold from 3D volumes, however is computationally expensive. The improvements are achieved in terms of processing speed (3X-10X speed-up depending on image size), minimizing sensitivity to initialization, and also increases local smoothness of the recovered manifold resulting in better reconstructed 2D composite image. We compare the proposed FastSME against the baseline SME as well as other accessible state-of-theart tools on synthetic and real microscopy data. Our evaluation on multiple metrics demonstrates the efficiency of the presented method in maintaining fidelity of manifold shape and hence specimen morphology.

LocalZProjector and DeProj: A toolbox for local 2D projection and accurate morphometrics of large 3D microscopy images

“…We generated the user-expected 2D projection of the epithelium (ground truth) by manually selecting the junctional signal of epithelial cells in each plane (see Additional File 1: Supplemental Note 1 for details). We then calculated several metrics measuring the accuracy and performance of the projection generated by the LocalZProjector tool, and compared the results to 7 other methods (Additional File 1: Supplemental Note 2): the standard maximalintensity-projection (MIP), the StackFocuser tool (9), Surf-Cut (12), PreMosa (10), the Extended-Depth-of-Field (EDF) tool (11), the Mininum-Cost-Z-Surface (MinCostZ) approach of (13,14), implemented in (15) and the Smooth Manifold Extraction tool (8,16). We first assessed how close the resulting projection was to the ground-truth projection using the root-mean square error (RMSE).…”

“…Yet the quality of projection dictates the subsequent step in analysis, as demonstrated in Additional File 1: Fig S3d. Some of these tools have the advantage of being parameter-free (8,(13)(14)(15)(16). LocalZProjector takes another approach and requires several parameters to be tuned.…”

“…To address this limitation, several projection tools have been developed that aim at including in the projection only the signal coming from the tissue layer. Among them there is StackFocuser (9), PreMosa (10), Extended Depth of Field (EDF) (11), SurfCut (12), MinCostZSurface (13-15), the Smooth Manifold Extraction (SME) tool (8) and a new implementation of the latter: FastSME (16). In (17), authors also proposed an approach based on Deep-Learning for the projection along the Z-axis, but it requires a set of images along with their already computed projections for its training (see Supplementary Information for descriptions).…”

Full volume super-resolution imaging of thick mitotic spindle using 3D AO STED microscope

“…It has been shown that it preserves local spatial relationships accurately while, in contrast, the MIP projection mixes all the stack layers together and does not provide a reliable geometric interpretation of the sample [45]. We used the Fast SME algorithm for decreasing computation time of calculation of SME projection [46].…”

“…The colormap used for displaying images is cubehelix colormap [44], which maximises contrast and has the advantage of being easily converted to grayscale without loss of information. Smooth manifold extraction (SME) projection [45] was implemented in Matlab using the code provided by the authors of the FastSME [46].…”