MiCellAnnGELo2: manual annotation enters the Metaverse

Session

Poster Session

Authors

Judith Lutton (1), Edward Offord (1), Till Bretschneider (1)

Affiliations

1. University of Warwick

Keywords

Annotation, Virtual reality, 4D microscopy,

Abstract text

MiCellAnnGELo (Microscopy and Cell Annotation Graphical Experience and Labelling Tool) was first released in 2022 [1] and provides a means of interaction with and fast annotation of segmented surfaces from 4D microscopy data. Most applications for annotation rely on a 2D screen, which is not always reliable for annotating 3D data. Annotating over multiple frames adds further complications to this approach when working with dynamic biological samples. A key technology that enabled the immersive graphical experience provided by MiCellAnnGELo is the virtual reality (VR) headset. This software has already been used for the analysis of complex and dynamic cell surfaces [2] and to generate training data for machine learning [3].

In the latest version, MiCellAnnGELo2, this application has been extended to run on the Meta Quest 2, which gives the option of experiencing the immersive annotation environment without requiring a high-end VR headset and graphics card. In this presentation, we will explore this and the other changes implemented in MiCellAnnGELo2 and demonstrate how this software is an ideal tool for manual annotation and data exploration.

MiCellAnnGELo2 allows annotation of 4D colour surfaces obtained from microscopy images. Annotation may take the form of markers, where the user marks points on the cell surface, or label painting, where labels are painted onto the surface to provide training annotations for machine learning. The interface is designed for ease-of-use, allowing fast annotation of features on a series of surfaces.

This software provides the perfect VR environment to allow annotation of dynamic surfaces captured with light microscopy. An increasingly important application of this software is in creating training annotations for machine learning. Future and experimental extensions include volume rendering, surface editing, and interfaces to enhance research demonstrations to the broader public.

References

[1] Platt, A., Lutton, J. E., Offord, E., & Bretschneider, T. (2023). MiCellAnnGELo: annotate microscopy time series of complex cell surfaces with 3D virtual reality. Bioinformatics, 39(1), btad013.

[2] Lutton, J. E., Coker, H. L., Paschke, P., Munn, C. J., King, J. S., Bretschneider, T., & Kay, R. R. (2023). Formation and closure of macropinocytic cups in Dictyostelium. Current Biology, 33(15), 3083-3096.

[3] Offord, E., Lutton, J. E., & Bretschneider, T. (2023). Cell Membrane Feature Detection Using Graph Neural Networks. 2023 IEEE 20th International Symposium on Biomedical Imaging (ISBI), pp. 1-5, doi: 10.1109/ISBI53787.2023.10230695.