PROJECT SUMMARY/ABSTRACT
Behavioral analysis of individual cells that determines the structural phenotypes in response to the mechanical
stimuli or cell migration is important in organogenesis or tissue morphogenesis. Dynamic cell tracking, migration
trajectory monitoring, or temporal changes in cell shape or size are essential for modeling cellular interactions.
However, optical limits imposed by microscope objective lenses and light attenuation in tissue hinder high-
resolution evaluation of cell physiology on the dimension scale of μm to nm. Hence, several super-resolution
fluorescence modalities have been developed in recent years to enable nanoscale tissue characterization and
less damage to tissue histopathology. On the other hand, isotropic nanometer optical resolution for super-
resolution modalities is restricted to micron scale field-of-view (FOV). This physical limitation reflects in vivo
characterization up to cellular events only. Therefore, we propose to develop an advanced optical imaging
technique to achieve sub-cellular resolution, while providing user capability to modulate millimeter to centimeter
FOV for in vivo monitoring. Light-sheet fluorescent microscopy (LSFM) has emerged as a popular optical
sectioning modality in biomedical research, owing to rapid camera frame rates in conjunction with long working
distance excitation optics. Although LSFM provides intermediate-to-high resolution images, the resolution is
highly dependent on the confocal range of the excitation objective. To overcome this challenge, we propose to
apply the rolling shutter (RS) based-axially sweeping LSFM technique to de-couple the dependence of light sheet
FOV on excitation numerical aperture (NA). RS is a type of image capture in sCMOS camera that record the
frame line by line on an image sensor instead of capturing the entire frame all at once to improve signal-to-noise,
but it can create some unintended image distortions. Thus, we will incorporate the application of maximum
likelihood estimation as a post-acquisition restoration strategy to remove optical distortions introduced by RS
image acquisition to ensure isotropic lateral and axial nanometer resolution. After the acquisition, we will segment
and extract information on cell motion and morphology using a feature detector framework based on the Hessian
difference of Gaussian in combination with the watershed algorithm. The feature detector has been validated to
separate adjacent clustered cells in 3D undergoing rapid motion with high sensitivity. In addition, we will further
train our segmented cell images for deep-learning network and use for automatic segmentation and tracking.
The proposed microscopic technology will enable the quantitative characterization of cellular behavior effectively
in arbitrary FOV.
