LipoPose: Adapting Cellpose to Lipid Nanoparticle Segmentation

Semanti Basu, Peter Bajcsy, Thomas Cleveland, Manuel Carrasco, R. Bahar

2023

Abstract

The goal of this study is to precisely localize lipid nanoparticles (LNPs) from cryogenic electron microscopy (cryoEM) images. LNPs found in cryoEM images are characterized by nonuniform shapes with varying sizes and textures. Moreover, there is no publicly available training dataset for LNP segmentation/detection. Thus, accurate supervised localization must overcome the challenges posed by heterogeneity of LNPs and nonexistent large training datasets. We evaluate benchmarks in closely related areas such as particle-picking and cell-segmentation in the context of LNP localization. Our experimental results demonstrate that, of the benchmarks tested, Cellpose is the best suited to LNP localization. We further adapt Cellpose to segmentation of heterogenous particles of unknown size distribution by introducing a novel optimization pipeline to remove uncertainty in Cellpose’s inference diameter parameter selection. The overall workflow speeds up the process of manually annotating LNPs by approximately 5X.

Paper Citation

in Harvard Style

Basu S., Bajcsy P., Cleveland T., Carrasco M. and Bahar R. (2023). LipoPose: Adapting Cellpose to Lipid Nanoparticle Segmentation. In Proceedings of the 16th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2023) - Volume 2: BIOIMAGING; ISBN 978-989-758-631-6, SciTePress, pages 115-123. DOI: 10.5220/0011726800003414

in Bibtex Style

@conference{bioimaging23,
author={Semanti Basu and Peter Bajcsy and Thomas Cleveland and Manuel Carrasco and R. Bahar},
title={LipoPose: Adapting Cellpose to Lipid Nanoparticle Segmentation},
booktitle={Proceedings of the 16th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2023) - Volume 2: BIOIMAGING},
year={2023},
pages={115-123},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011726800003414},
isbn={978-989-758-631-6},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 16th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2023) - Volume 2: BIOIMAGING
TI - LipoPose: Adapting Cellpose to Lipid Nanoparticle Segmentation
SN - 978-989-758-631-6
AU - Basu S.
AU - Bajcsy P.
AU - Cleveland T.
AU - Carrasco M.
AU - Bahar R.
PY - 2023
SP - 115
EP - 123
DO - 10.5220/0011726800003414
PB - SciTePress