Speckle applications for volumetric imaging of biological dynamics

Abstract

Speckle is a high contrast granular pattern formed by coherent light interactions with rough surfaces or biological tissue. The speckle intensity fluctuates in space (or time) in a "random" manner and this randomness is often treated as noise in optical imaging. In another perspective, the speckle statistics can be exploited to extract information that is not readily accessible through conventional imaging techniques. This thesis describes the application of speckle in assisting imaging structure and dynamics of various biological processes, including laser speckle contrast imaging (LSCI) for blood flow measurements and HiLo microscopy with speckle illumination to map neuronal activities.

For LSCI, quantitative analysis that converts contrast measurement to speckle correlation time constant is important when monitoring blood flow changes. We describe a direct integration method based on temporal speckle analysis to estimate the speckle correlation time. This method enables a model-free association of the contrast to the speckle field dynamics in cases of ergodic measurements and does not require numerical fitting. We also present approaches to correct errors from multiple sources in speckle contrast measurements, including sampling bias due to limited statistics and detection noise from shot noise and readout noise.

Next, we describe HiLo microscopy with speckle illumination, which is a previously developed wide field technique in fluorescence imaging with optical sectioning capabilities. Speckle HiLo is easy to implement and robust to aberrations in the illumination path, but suffers from residual speckle artefact in the final reconstruction. We propose a non-local means based denoising method tailored for HiLo microscopy to reduce speckle noise, which computationally mitigates the trade-off between image fidelity and sectioning strength.

Finally, multifocus imaging is presented in both speckle imaging modalities, enabling single shot volumetric imaging. This is achieved by introducing a passive optical element called z-splitter prism, through which images from multiple depths can be projected onto a single camera simultaneously. With the z-splitter, both multifocus LSCI and HiLo systems are advantageous in high speed recording as compared to conventional systems requiring axial scan.