DataDirect Networks, a provider storage technology for HPC and enterprise, has announced a breakthrough in microscopy workflows that enables real-time analysis for improved resolution of images.
Comprehensive testing of DDN AI200 shared parallel storage appliance, a NVIDIA DGX-1 system and the Microvolution advanced processing software demonstrates significant experimental and economic benefits, delivering 1600-fold better yield compared with traditional microscopy workflows, and 8X faster time to insight than using cloud-based infrastructure.
Lattice LightSheet microscopes and other high-data-rate instruments are revolutionising biological research in areas such as neuroscience, developmental biology, cancer research and biomedical engineering. To enable the incredible resolution of these high-end instruments, researchers employ sophisticated deconvolution algorithms to increase signal-to-noise and to remove blurring in these 3D images. However, researchers have historically faced efficiency challenges when storing, managing and processing the large amounts of data generated, and the sheer amount of processing required limited these techniques to a post-processing step rather than being applicable in real-time during imaging capture.
‘Deconvolution has historically been time-consuming and resource-intensive,’ said Marc Bruce, CEO of Microvolution. ‘Traditional methods require significant compute resources running for long periods of time to process a single image. The lack of real-time feedback from the images reduces the efficiency of the microscopy pipeline and limits its usefulness. This ground-breaking platform enables real-time analysis and live rendering of experiments, providing immediate feedback to the microscope operator to ensure running under optimal conditions.’
‘Real-time analysis and imaging provide unprecedented flexibility for experiments,’ said George Vacek, global director for life sciences, healthcare and machine learning at DDN. ‘The platform critically enables the increased temporal resolution that these instruments offer. Efficient detection allows for imaging of dim and fast phenomenon that were not previously possible to resolve.’
Featuring a DDN AI200 shared parallel storage appliance, a NVIDIA DGX-1 server and the Microvolution advanced processing software, the accelerated microscopy platform enables and accelerates microscopy workflows for deconvolution and deskewing, showing real-time capability even for the highest data rate instruments. It enables more efficient utilisation of shared instruments, provides unlimited data capture from multiple instruments and simplifies workflows by eliminating the need for time-consuming, error-inducing data management tasks. It also includes robust data protection and access control mechanisms to meet data governance and regulatory compliance requirements. Configured for a small initial footprint, the solution scales easily in capacity, performance and capability to match evolving workflow requirements and future investments in instruments.