Year: 2023
Type: conference-abstract
Abstract: Many biological phenomena are challenging to study in part because they are inherently rare. For example, in the bone marrow, 1 in 4000 cells are osteogenic progenitors [1, 2]. Indeed, owing to its combinatorial nature, this problem is further confounded when one wishes to characterize higher order events such as heterotypic cell-cell interactions or identify cells positive for multiple recombination events under sparse labeling conditions. To evaluate such processes in large, chemically cleared multicellular contexts, we developed Multiscale Cleared Tissue Axially Swept Light-Sheet Microscope (MCT-ASLM) [3-6]. MCT-ASLM consists of two imaging systems, referred to as the macroscale and nanoscale modules, arranged in a 4-axis geometry. The macroscale module is a single-sided illumination variant of mesoSPIM [7]. Here, illumination is delivered to the specimen with a modified Nikon AF-S 50mm camera lens in a digitally scanned light-sheet format for shadow reduction. An electro-tunable lens, conjugate to the back pupil of the camera lens, scans the laser beam in its propagation direction synchronously with the rolling shutter of a CMOS camera, thereby enabling low numerical aperture ASLM. Detection is performed with a variable magnification Olympus MVX-10 macro microscope and a Hamamatsu Flash 4.0 V3 CMOS camera. The nanoscale module is constructed with NA 0.7 multi-immersion illumination and detection objectives [6]. Owing to their design, these objectives have a large working distance (10mm) and deliver spherical and chromatic aberration free imaging in diverse solvents with refractive indices between 1.33 and 1.56. Aberration-free remote focusing is carried out with a pneumatically driven voice coil, and shadow reduction provided with a resonant galvanometer conjugate to the specimen in a multidirectional illumination format [8, 9]. The specimen is suspended at the center of the macroscale and nanoscale modules from large travel distance (∼50-200 mm) precision translation and rotation stages and switching between the imaging modes is rapid. MCT-ASLM is controlled by custom Python software organized in an industry-standard Model-View-Controller architecture. To maximize performance, it extensively leverages threading, parallelization, and passes large objects as pointers between different processes. Image analysis is implemented in a hybrid CPU/GPU format, and includes classical, machine-learning, and deep-learning-based feature extraction. This includes a REST-API layer for communication with external python packages such as ilastik [10]. All imaging and analysis steps are executed as a reconfigurable Tree Data Structure, thereby allowing the creation of acquisition ‘recipes’ according to the specimen. Accordingly, the microscope interrogates biological features with varying degrees of automation, including manual, human-in-the-loop, and fully autonomous modes. Thus, MCT-ASLM is capable of rapidly imaging large (∼20 x 20 x 20mm) cleared tissues with 5-micron resolution, automatically identifying biological features of interest with cutting-edge computer vision, and subsequently interrogating these events with an isotropic resolution of 300nm.
Source: Microscopy and Microanalysis
Institution: The University of Texas Southwestern Medical Center
Language: English
Subfield: Biophysics
Domain: Life Sciences
Open Access status: bronze
OpenAlex ID: W4385071946