Boston University
Advancing single-cell analysis: deep learning for cell segmentation and tracking with improved metrics
Abstract
dc:description.abstractAdvances in microscopy software and hardware have dramatically accelerated the pace of image acquisition, making analysis a major bottleneck in generating quantitative, single-cell data. To address this challenge, we leverage deep learning to develop robust software for cell segmentation and tracking. First, we present DeLTA 2.0, a convolutional neural network-based pipeline optimized for high throughput analysis of single-cell images on two-dimensional surfaces, enabling quantification of gene expression and cell growth. Next, we introduce Cell-TRACTR, a transformer-based model that integrates segmentation and tracking into a unified framework, delivering enhanced tracking accuracy. Lastly, we propose Cell-HOTA, a novel cell tracking metric that balances detection, association and division accuracy, offering a balanced and interpretable evaluation of cell tracking performance. Collectively, these contributions advance the field of cell tracking, enabling more accurate and robust single-cell data analysis.
Degree
thesis:*- Name thesis:degree_name
- Doctor of Philosophy
- Level thesis:degree_level
- doctoral
- Discipline thesis:degree_discipline
- Biomedical Engineering
- Grantor
- Boston University
- Year dc:date.issued
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- O'Connor, Owen M.
- Advisor dc:contributor.advisor
-
- Dunlop, Mary J.
- ORCID dc:identifier.orcid
Subjects
dc:subject × 4Rights
dc:rights- Statement dc:rights
-
- Attribution 4.0 International
- Licence dc:rights.uri
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/2144/50613