Stitching with software
Novel methods of optical microscopy yield highly detailed images of whole organs – and enormous amounts of data. A new program called BigStitcher now enables the big picture to be pieced together with no loss of fine detail.
The amount of detail recorded in these images is astonishing. They reveal the tiniest structures present in each and every cell, yet whole organs can also be captured with this amazing level of precision. Meanwhile, new chemical methods of tissue preparation allow whole organs to be rendered transparent, and these can be imaged by light-sheet microscopy – one of the many different approaches to high-resolution optical microscopy that have emerged in recent years.
But in order to analyze the finest details of subcellular structures in the context of whole organs or indeed organisms, the connections between single neurons in the whole brain of a mouse, for example – one needs to find a way of assembling many single images to a faithful picture of the whole. Data scientists refer to this task as stitching. The wealth of detail encoded in these images often amounts to several terabytes (TB), which is equivalent to the pictorial content of several million photographs. Processing of such huge masses of data poses a tremendous challenge for stitching software, whose job is to ‘line up’ all the single images automatically, comprehensively and seamlessly.
Now, researchers led by Professor Heinrich Leonhardt (Department of Biology II, LMU) and Dr. Stephan Preibisch (Max Delbrück Center for Molecular Medicine, Berlin), in collaboration with partners based at the Howard Hughes Medical Institute’s Janelia Research Campus in Ashburn, Virginia (USA), have now developed the software package BigStitcher, which marks a significant advance in the technology available for the faithful processing of high-resolution imaging data. The new program provides a wide selection of functions which facilitate the management of these high-dimensional datasets, and allow users to efficiently assemble many single, three-dimensional images into a coherent whole. In addition, the new software can correct for various types of optical errors, such as chromatic aberration or local geometrical distortions.
A major part of the development effort was devoted to ensuring ease of use and compatibility with conventional hardware. “Very few research groups and microscopy centers employ dedicated specialists in image analysis,” says David Hörl, first author on the new paper and a PhD student in Heinrich Leonhardt’s lab. “With our new software, we hope to make it possible for users who don’t have advanced knowledge of informatics to process large image files successfully. BigStitcher’s graphical user interface (GUI) provides a live preview of the results, so that any problems that arise in the various processing steps can be recognized early on. This avoids having to repeat the whole procedure from scratch. Thanks to clever management of the data in multiresolution pyramids, comparable to the approach used by Google Maps, the program allows very large image datasets to be stitched on ordinary computers.” The researchers have developed BigStitcher as a plug-in for the popular image-processing program ImageJ. The source code is freely available.
Nature Methods 2019