'BigStitcher' reconstructs terabyte datasets


Rebecca Pool

Wednesday, August 14, 2019 - 21:45
Image: Cross-section through the complete nervous system of a fruit fly larva. [Janelia/MDC]
A team of researchers from Germany, UK and the US have unveiled software to process the vast swathes of data generated by expansion microscopy and light sheet imaging.
Light-sheet imaging of cleared and expanded samples creates terabyte-sized datasets that consist of many unaligned 3D image tiles.
So-called BigStitcher can reconstruct these before analysis, enabling interactive visualisation, fast and precise alignment, deconvolution of multi-view datasets and much more.
“With the aid of modern light-sheet microscopes, which are now found in many labs, large samples processed by these methods can be rapidly imaged,” says Dr Stephan Preibisch from the Berlin Institute for Medical Systems Biology.
“The problem, however, is that the procedure generates such large quantities of data - several terabytes - that researchers often struggle to sift through and organise the data,” he adds.
To tackle this problem Preibisch and colleagues including researchers from Janelia Research Campus, Howard Hughes Medical Institute, US, developed BigStitcher, which delivers reconstructions that resemble, as they say, Google Maps in 3D mode.
“One can not only get an overview of the big picture, but can also zoom in to specifically examine individual structures at the desired resolution,” highlights Preibisch.
“Our software runs on any standard computer,” he adds. “This allows the data to be easily shared across research teams.“  
Two virtual right-angled cross-sections through the complete nervous system of a fruit fly larva. Each section highlighted in colour is a large 3D image that has been automatically assembled like a mosaic into an overall image hundreds of gigabytes in size. [Janelia / MDC]
As the researchers explain in Nature Methods, BigStitcher enables interactive visualisation, fast and precise alignment, spatially resolved quality estimation, real-time fusion and deconvolution of dual-illumination, multitile, multiview datasets.
What's more, the software also compensates for optical effects, thereby improving accuracy and enabling subsequent biological analysis.
The program is distributed within the Fiji framework, where any interested scientist can download and use the plug-in free of charge.
Research is published in Nature Methods.
Website developed by S8080 Digital Media