Researchers at the Moffitt Cancer Center have created a new open-source software program that allows users to view multiple 2D images at once. Their program, called Mistic, extracts information from multidimensional images to create a summary of each that can be viewed together. A paper in the news Patterns describes the program and various applications in cancer imaging.
“There are many commercial and open-source imaging technologies that you can use to visualize and process just one of those images,” explains lead author Sandhya Prabhakaran, PhD, lead author and applied researcher at Moffitt. Users can open those programs to open an image to get a better understanding or deeper insights of one image. “We wanted to create a software program where we didn’t have to click on each photo or use additional software to view the images,” adds Prabhakaran. “Instead, we wanted to group multiple images together to see what pattern exists.” With Mistic, users can pull in multiples of these high-dimensional images created via Vectra, CyCIF, t-CyCIF, and CODEX and then view them all together.
After looking at other image-editing tools, the team realized they needed a better method of viewing both multiplexed and single images, and that they had to run on standard Mac computers. None of the tools available provided all of these necessities, and many were not intended for biomedical use. “We realized that we needed to build our own software so that we could see spatial patterns between the tumor and the immune cells we were studying,” explains Prabhakaran.
Mistic builds on recent landmark improvements in imaging technologies for studying tissue samples. For example, machines can now be programmed to stain hundreds of slides simultaneously, or alternatively, up to 1,000 different tissue sample cores can be placed on a single slide and stained for biomarkers simultaneously. With the advent of these approaches comes a wealth of opportunities to generate new data and information. Due to the magnitude of this information and the complex nature of cancer itself, computational modeling and software are needed to view and study the cancer biomarkers, tissue architecture, and cellular interactions between these samples.
Mistic relies on dimensionality reduction – a process of reducing high dimensional data to lower dimensionality to simultaneously create a bird’s eye view of each image. “Then you can use Mistic to select a particular image that you want to study further with advanced processing software, such as Fiji or QuPath,” Prabhakaran said. “We didn’t want to reinvent the wheel with all this high-performing, well-coded software, so we think of Mistic as a first pre-processing data visualization tool that gives the user a first insight into what all these images look like. they can then choose which images to choose for downstream processing.”
The researchers argue that Mistic can be used for a variety of purposes, including identifying biomarkers and understanding tissue architecture and the spatial organization of different cell types. For example, the researchers showed that the software can be used to view 92 images of patients with non-small cell lung cancer and infer how biomarkers cluster between patients with different responses to treatment. In another example, the researchers used Mistic in conjunction with statistical analysis to assess the spatial colocalization and co-expression of immune cell markers in 210 endometrial cancer samples.
“With this open-source software, users can now view all of your images simultaneously — whether for individual biomarkers or multiplex images — and for multiple cancer types,” added senior author Alexander Anderson, PhD, chair of the Division of Integrated Mathematical Oncology at Moffitt. “And we certainly see that it can be applied to all areas and domains, not necessarily just for cancer imaging.”