Imaging across multiple spatial scales with the multi-camera array microscope

Abstract

This paper experimentally examines different configurations of a multi-camera array microscope (MCAM) imaging technology. The MCAM is based upon a densely packed array of “micro-cameras” to jointly image across a large field-of-view (FOV) at high resolution. Each micro-camera within the array images a unique area of a sample of interest, and then all acquired data with 54 micro-cameras are digitally combined into composite frames, whose total pixel counts significantly exceed the pixel counts of standard microscope systems. We present results from three unique MCAM configurations for different use cases. First, we demonstrate a configuration that simultaneously images and estimates the 3D object depth across a 100×135mm2 FOV at approximately 20 µm resolution, which results in 0.15 gigapixels (GP) per snapshot. Second, we demonstrate an MCAM configuration that records video across a continuous 83×123mm2 FOV with twofold increased resolution (0.48 GP per frame). Finally, we report a third high-resolution configuration (2 µm resolution) that can rapidly produce 9.8 GP composites of large histopathology specimens.

High-throughput imaging across multiple spatial scales with the multi-camera array microscope (MCAM)

“…The primary design of the MCAM includes 54 micro-cameras that are packed on a grid with 13.5mm centerto-center spacing. 5 Other designs include up to 96 cameras. 6 Beyond its obvious advantage to simultaneously image a large FOV at high resolution, the MCAM allows different configurations for different functionalities.…”

“…6 Beyond its obvious advantage to simultaneously image a large FOV at high resolution, the MCAM allows different configurations for different functionalities. 5 Multi-view imaging for 3D reconstructions In this configuration, the MCAM uses lenses with a large working distance and a low image magnification. Thereby, the FOVs of individual lens-sensor pairs overlap and each location is viewed by multiple cameras at different angles.…”

“…Thereby, the FOVs of individual lens-sensor pairs overlap and each location is viewed by multiple cameras at different angles. This configuration allows to capture 3D snapshots across a 100x135mm² area at approximately 20 µm full-pitch lateral resolution and 42 µm axial sensitivity using 54 cameras 5 or across a FOV of 135 cm² at up to 230 frames per second (total throughput of 5 gigapixels per second). 7 The reconstruction algorithm stitches all 54 images into a composite and generates a co-registered 3D height map.…”

“…7 The reconstruction algorithm stitches all 54 images into a composite and generates a co-registered 3D height map. 5,7 Thereby, the height information is based on the stereo overlap. A self-supervised neural network is trained to map raw images to 3D topography and only uses stereo overlap and ray-propagation physics as supervision mechanism.…”

“…This configuration allows to capture video across a continuous 83x123mm² area at 2X higher lateral resolution with 54 cameras. 5 Tiled imaging for high-resolution microscopy This configurations aims at microscopic resolution and therefore uses higher magnification. Thus, the tiled configuration no longer images across a continuous area in one snapshot.…”