Computer vision tools to automate the study of Schlemm's Canal inner wall cell biomechanics and pore formation
File(s)
Author(s)
Rodrigues, Justino
Type
Thesis
Abstract
Glaucoma is the world's leading cause of irreversible blindness and is associated with elevated intraocular pressure, which is often elevated due to an increase in aqueous humor outflow resistance. Aqueous humor outflow resistance is thought to be modulated by Schlemm’s canal inner wall pores, which form in response to cellular deformation. In glaucoma, pore density is decreased, suggesting that impaired pore formation may contribute to increased outflow resistance and elevated intraocular pressure in glaucoma. Currently available methods of studying these pores have significant limitations and thus our understanding of Schlemm's canal pores remains rudimentary. This thesis aims to provide tools that address the limitations of existing techniques and to provide new techniques which allow for deeper understanding of Schlemm's canal pores.
The first chapter provides a machine-learning system for the detection of pores in previously obtained image data. The second chapter uses machine learning tools to quantify pore density and size, showing that despite controlling cellular deformation, pore formation is reduced in glaucoma, and that pore size also increases with stretch and is reduced in glaucoma. The third chapter presents an algorithm for tracking the mechanical strain in cells in real time and a modification to an existing method of studying Schlemm's canal pores which allows for the investigation of time dependent phenomena, including preliminary evidence that Schlemm's canal cells retain their pore formation ability in subconfluent culture and that Schlemm's canal pores are transient structures.
Together, the presented methods augment the ability to research Schlemm's canal pores by decreasing research time and allowing for new aspects of Schlemm's canal pores to be explored. The findings presented indicate that despite controlling for cell deformation, pore formation ability is still reduced in glaucoma. This suggests that cell stiffness is not the sole reason for the decreased pore density seen in glaucoma.
The first chapter provides a machine-learning system for the detection of pores in previously obtained image data. The second chapter uses machine learning tools to quantify pore density and size, showing that despite controlling cellular deformation, pore formation is reduced in glaucoma, and that pore size also increases with stretch and is reduced in glaucoma. The third chapter presents an algorithm for tracking the mechanical strain in cells in real time and a modification to an existing method of studying Schlemm's canal pores which allows for the investigation of time dependent phenomena, including preliminary evidence that Schlemm's canal cells retain their pore formation ability in subconfluent culture and that Schlemm's canal pores are transient structures.
Together, the presented methods augment the ability to research Schlemm's canal pores by decreasing research time and allowing for new aspects of Schlemm's canal pores to be explored. The findings presented indicate that despite controlling for cell deformation, pore formation ability is still reduced in glaucoma. This suggests that cell stiffness is not the sole reason for the decreased pore density seen in glaucoma.
Version
Open Access
Date Issued
2022-07-21
Date Awarded
01/04/2023
Advisor
Overby, Darryl
Sponsor
Imperial College London
Maples Group (Firm)
Publisher Department
Bioengineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)