Deep learning and mathematical analysis of wound healing in flies

: Live imaging in Drosophila

Abstract

Wound healing is a highly conserved process required for survival after tissue damage. In mammals, the three cell-behaviours that contribute to wound re-epithelialisation are cell shape deformation, cell division, and cell migration. This study aims to quantify the contributions of each of these cell behaviours using wounded Drosophila pupae. Live confocal time-lapse microscopy allows us to follow cell behaviours in the pupal wing before and after wounding. We have developed deep learning algorithms to identify dividing cells with 97% accuracy and determine the orientation of these divisions relative to the wound margin, alongside additional machine learning tools that measure other cell behaviours. We want to know whether these properties are synchronized/aligned by the chemical and mechanical signals of wounding and have developed a linear continuum model of the system to better understand the interplay between the contributing cell behaviours. We are also investigating how far back from the wound edge these changes in cell behaviour occur, and whether this depends on wound size. We have characterised the spatial-temporal distribution of divisions and found a reduction in their density close to the wound edge, but 2hr after wounding there is a synchronised burst of divisions further back. Next, we will genetically modify wound signals to knockdown one or more cell behaviours and examine not only the gross effect on wound closure, but also how the other behaviours compensate. Having a greater understanding of the mechanics of wound healing will hopefully be a first step in developing future treatments for the clinic.

Date of Award	20 Sept 2023
Original language	English
Awarding Institution	University of Bristol
Supervisor	Isaac V Chenchiah (Supervisor), Helen M A Weavers (Supervisor), Tanniemola B Liverpool (Supervisor) & Paul B Martin (Supervisor)