Back to search

University of British Columbia

Development of a machine vision guided inkjet platform for single cell isolation using real-time machine learning feedback

Abstract

dc:description

Single cell level analysis has become an invaluable tool for researchers, enabling investigations into cellular heterogeneity across genomic, transcriptomic, and proteomic modalities. Unlike bulk analysis methods that average signals across large populations, single cell investigations can offer insights into rare or functionally distinct subpopulations that drive critical biological processes, such as tumor progression or therapeutic resistance. This thesis presents the development of a novel cell handling platform that utilizes inkjet printhead technology integrated with machine vision feedback guided by a trained neural network classifier to enable precise isolation and deposition of individual cells. The custom neural networks were trained on a diverse set of manually annotated cell lines and demonstrated strong generalization to previously unseen cell types, enabling robust and flexible single cell capture. The small size of the neural networks allowed for low-latency inference of an average of 9.72 ms per image, facilitating real-time feedback. The system is designed to improve throughput and accuracy in sample preparation via a novel reactive variable volume print strategy, by printing over a target until a cell dispensing event is observed, addressing key limitations in capture efficiency and throughput in current single cell systems. The deployed neural network was benchmarked at a weighted-F1 score of 0.986, achieving 71.2% single cell occupancy at a throughput of 0.52 seconds per dispensed single cell. Built in data traceability ensures high confidence in sample purity, providing a robust tool for adapting established workflows in addition to the development of novel assays. The instrument supports common open well substrate formats, including the development of a dew point tracking substrate chiller which maintains volumetric accuracy of the low volume reactions in nanowell chips. The utility of the instrument was validated using established single cell genomics and transcriptomics protocols: DLP+ and VASA-Seq. Furthermore, a novel single cell proteomics workflow was developed, leveraging the instrument’s capabilities, highlighting the flexibility of the open platform in handling sensitive protein signals. By facilitating consistent and efficient preparation of single cell samples, this work contributes to the advancement of high-throughput single cell omics studies, supporting deeper insight into complex biological systems and disease mechanisms.

Degree

thesis:*
Name thesis:degree_name
Doctor of Philosophy - PhD
Level thesis:degree_level
doctoral
Discipline thesis:degree_discipline
Electrical and Computer Engineering
Grantor dc:publisher
University of British Columbia
Year dc:date
2025

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Cheng, Eric
Contributors dc:contributor
  • Cheung, Karen C.

Rights

dc:rights
Statement dc:rights
  • Attribution-NonCommercial-NoDerivatives 4.0 International
Language dc:language
eng

Identifiers

dc:identifier.*
Handle dc:identifier
http://hdl.handle.net/2429/92181

Chain of custody

source
Harvested from
University of British Columbia
Base URL
circle.library.ubc.ca/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Cheng, Eric. Development of a machine vision guided inkjet platform for single cell isolation using real-time machine learning feedback. doctoral thesis, University of British Columbia, 2025. http://hdl.handle.net/2429/92181