GroundTruthVR: Identifying Cell Membrane Ground Truth For Automated Image Segmentation in Virtual Reality

Assigned Computer Science Games Engineering Bachelor Master Thesis Games Research Lab

Background

The vast datasets created by microscopy imaging are usually processed by automated analysis tools to identify rel- evant features, such as different tissue types, cell nuclei, or other structural and functional features. However, each automated analytical step needs to be validated and benchmarked. For this, scientists painstakingly markup ground truth data as a validation set. The GroundTruthXR project aims to support the mark-up process of volumetric datasets using immersive visualization tools and an intuitive user experience. Thin cell membranes are prone to image artifacts that can decrease the effectiveness of automated segmentation algorithms. The thesis focuses on the manual markup of cell membranes from images of pre-implantation embryos using virtual reality (VR) hardware. It provides an overview of the current state of the art regarding automated segmentation, with a focus on cell mem- branes. The main contributions are a concept for the definition of cell membranes using an immersive application. This includes a formal description of the process, and the description of individual virtual tools and interaction methods required to perform necessary steps. The prototype should be developed following user-centered design principles and subjected to frequent review by domain specialists. To evaluate the effectiveness and user response, a basic comparison with traditional desktop-based methods could be performed.

Tasks

1. Review of the state of the art in identifying cell membranes.
   • Including the current workflow for generating ground truth data on cell membranes.
2. Conceptualise a new workflow for defining ground truth data on cell membranes using virtual reality hardware.
   • Identify suitable methods of describing cell membranes in 3D, based on the available data and the requirements for validating the results of automated algorithms.
   • Identify algorithms for modifying such membrane descriptions.
   • Design virtual tools to allow the user to work with membranes in an immersive spatial environment, i.e. manage and model membranes.
3. Implement and validate a prototype based on the above using an iterative, user-centered design.
   • The prototype must accept real-world data sets and provide the necessary features for adjusting the visualisa- tion.
   • The prototype should integrate with the previous work within the GroundTruthXR project.

Prerequisites

• Interest in biology
• Good knowledge of C#

Contact Persons at the University Würzburg

References

• Beyer, Johanna, Markus Hadwiger, and Hanspeter Pfister. 2015. “State-of-the-Art in GPU-Based Large-Scale Volume Visualization.” Comput. Graph. Forum 34 (8): 13–37. https://doi.org/10.1111/cgf.12605.
• 1
• Ganovelli, Massimiliano AND Pattanaik, Fabio AND Corsini. 2014. Introduction to Computer Graphics - a Practical Learning Approach. Boca Raton, Fla: CRC Press. https://learning.oreilly.com/library/view/introduction-to- computer/9781439852798/.
• LaViola, J. J., E. Kruijff, R. P. McMahan, D. A. Bowman, and I. Poupyrev. 2017. 3d User Interfaces: Theory and Practice. Addison-Wesley Usability and HCI Series. Addison-Wesley. https://books.google.de/books?id=ilUyjw EACAAJ.
• Mathew, B., A. Schmitz, S. Muñoz-Descalzo, N. Ansari, F. Pampaloni, E. H. K. Stelzer, and S. C. Fischer. 2015. “Robust and Automated Three-Dimensional Segmentation of Densely Packed Cell Nuclei in Different Biological Specimens with Lines-of-Sight Decomposition.” BMC Bioinformatics 16 (1). https://doi.org/10.1186/s12859-015- 0617-x.
• Roeder, A. H. K., A. Cunha, M. C. Burl, and E. M. Meyerowitz. 2012. “A Computational Image Analysis Glossary for Biologists.” Development 139 (17): 3071–80. https://doi.org/10.1242/dev.076414.