Title:	Deep learning based species identification of bacteria using time-lapse of growth in microfluidic chip imaged by phase contrast microscopy
DNr:	Berzelius-2025-366
Project Type:	LiU Berzelius
Principal Investigator:	Carolina Wählby <carolina.wahlby@it.uu.se>
Affiliation:	Uppsala universitet
Duration:	2025-10-24 – 2026-05-01
Classification:	10207
Homepage:	https://strategiska.se/en/research/ongoing-research/ssf-agenda-2030-research-centers-arc-2019/project/10866/
Keywords:

Abstract

In this project, we are investigating whether it is possible to train deep learning models to identify bacterial species growing in a microfluidic trap based on their spatiotemporal division patterns. This has important implications for selecting appropriate antibiotics, reducing the use of broad-spectrum antibiotics, and improving patient outcomes. We have previously demonstrated this approach using laboratory isolates in several publication. In this final year, we aim to extend the method to clinical patient isolates. Berzelius has been properly acknowledged in all our previous publication (mentioned in the activity reports)