Biomedical applications of holographic microscopy

Abstract

Identifying patients with aggressive cancers is a major healthcare challenge in resource-limited settings such as sub-Saharan Africa. Holographic imaging techniques have been shown to perform diagnostic screening at low cost in order to meet this clinical need, however the computational and logistical challenges involved in deploying such systems are manifold. This thesis aims to make two specific contributions to the field of point-of-care diagnostics. First, it documents the design and construction of low-cost holographic imaging hardware which can serve as a template for future research and development. Second, it presents a novel deep-learning architecture that can potentially lower the computational burden of digital holography by replacing existing image reconstruction methods. We demonstrate the effectiveness of the algorithm by reconstructing biological samples and quantifying their structural similarity relative to spatial deconvolution methods. The approaches explored in this work could enable a standalone holographic platform that is capable of efficiently performing diagnostic screening at the point of care.