Taking Advantage of Advances in Computer Hardware for Image Quantification

Here at Visikol, image analysis is a central pillar of our experimental process, and how we turn complex in vitro disease models into actionable insights for addressing specific biological questions. When it comes to image analysis, we are typically more focused on the software side of things as we have developed a large suite of image analysis software for use with two-dimensional data sets (e.g. slide scanning) as well as three-dimensional multiplex data sets (e.g. confocal or light sheet microscopy). Our software suite extends from traditional threshold-based image quantification software to answer discrete biological questions all the way to bespoke machine learning solutions which can be deployed for Clients on site. We currently only leverage this software internally for our services and thus receive a lot of questions from researchers using our 3D imaging reagents about how to distill their terabytes of multi-channel imaging data into useful insights.

Over the last few years there have been recent hardware developments that will have large implications for everyone taking advantage of image processing in their drug discovery pipeline that merits some discussion.

Deep learning neural networks rely on the computational power of GPUs, while traditional image quantification leans much more heavily on CPUs and RAM, and CPUs are the topic for this post. AMD is currently releasing their 3rd Generation Ryzen Chipset based on the 7 nm process node and Zen 2 architecture, a large improvement over existing hardware. The top consumer chip has 16 multithreaded cores with ECC support. This is significant because it gives the consumer access to both significantly higher core count CPUs, and high-speed individual core performance. This was something that was previously only available to those with access to high performance computer clusters.

CellProfiler and ImageJ, are both publicly available open source software packages for measuring and analyzing biological images whose performance scales well with access to additional CPU cores. This expansion in hardware availability and computational power will give more researchers access to powerful image processing tools and we are excited to see the results of this hardware democratization.

If you have a large image set that you would like advice on how to quantify, or a complex biological problem that you think deep learning can solve, please reach out to a Visikol Scientist today.

Contact our research team to learn more


Share This Page, Choose Your Platform!