High performance computing for high throughput bioimaging

Session

Poster Session

Authors

Victor Ionescu (1), Marie Held (1), Marco Marcello (1), Alison Beckett (1), Andrew Collins (1)

Affiliations

1. Liverpool Shared Research Facilities, University of Liverpool

Keywords

HPC, segmentation, denoising, lsm, sem

Abstract text

Data processing and analysis has been identified as the area which typically consumes the most time for bioimaging researchers and requires the most involvement (Schmidt, C., et al. 2022). In the fields of electron and light microscopy this can be in part due to the large volume of data produced by techniques such as Scanning/Volume Electron Microscopy (SEM/vEM) and Laser Scanning Microscopy (LSM). The analysis of this type of data can require a high level of prior knowledge and the complexity of tools to analyse this data can limit automation to only the most experienced image analysts, with others preferring to process large volumes of data by hand. In recent years a number of tools have become available that leverage machine learning removing the a priori requirement, such as Cellpose (Pachitariu, M. et al., 2022), Deconwolf (Wernersson, E., et al., 2022) and CLEM-Reg (Krentzel, D et al., 2023). Despite the high computational requirements of these tools and high volume of data, many are still used through a GUI on a single workstation. In this work we have developed workflows for multiple common bio-imaging tools to allow them to utilise the full capabilities of a High-Performance Computing (HPC) platform providing significant reductions in overall run time. Further, in this work we show how these workflows can be integrated into a GUI so that these tools can be accessed with minimal command line experience.

References

Schmidt, C., Hanne, J., Moore, J., Meesters, C., Ferrando-May, E. and Weidtkamp-Peters, S., 2022. Research data management for bioimaging: the 2021 NFDI4BIOIMAGE community survey. F1000Research, 11.

Krull, A., Buchholz, T.O. and Jug, F., 2019. Noise2void-learning denoising from single noisy images. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 2129-2137).

Pachitariu, M. and Stringer, C., 2022. Cellpose 2.0: how to train your own model. Nature methods, 19(12), pp.1634-1641.

Wernersson, E., Gelali, E., Girelli, G., Wang, S., Castillo, D., Langseth, C.M., Nguyen, H., Chattoraj, S., Casals, A.M., Lundberg, E. and Nilsson, M., 2022. Deconwolf enables high-performance deconvolution of widefield fluorescence microscopy images.

Krentzel, D., Elphick, M., Domart, M.C., Peddie, C.J., Laine, R.F., Henriques, R., Collinson, L.M. and Jones, M.L., 2023. CLEM-Reg: An automated point cloud based registration algorithm for correlative light and volume electron microscopy. bioRxiv, pp.2023-05.