Microscopy samples are seldom completely flat across a centimeter-scale field of view. Mechanical scanning can keep all the parts of a large sample in focus, but scanning reduces throughput, slowing the imaging process. To help large-area microscopy systems resolve trade-offs between field of view, resolution, and imaging speed, a team at Duke University developed a single-shot, re-imaging microscope that achieves seamless, gigapixel imaging over a 16.3 x 18.8 square millimeter (mm2) field of view, at 0.84-µm half-pitch resolution, without mechanical scanning. A close-up view of the PANORAMA microscope’s camera array, which contains 48 tiny cameras that work together to capture ultrahigh-resolution, gigapixel images of large and non-flat objects in a single snapshot. Courtesy of Duke University/Roarke Horstmeyer. The microscope, which the researchers call PANORAMA, could enhance imaging applications for biological research and medical diagnostics, as well for industrial inspection and quality control. “This tool can be used wherever large-area, detailed imaging is needed,” researcher Haitao Chen said. “For instance, in medical pathology, it could scan entire tissue slides, such as those from a biopsy, at cellular resolution almost instantly. In materials science or industrial inspection, it could quickly inspect large surfaces, such as a chip wafer, at high detail.” PANORAMA uses a telecentric photolithography lens, a large-aperture tube lens, and a flat micro-camera array with adaptive, per-camera focus control to provide sub-µm focus across flat, curved, and uneven samples that span cm. The telecentric lens, originally developed for chip-making, is combined with a large tube lens that projects an image of the sample onto a flat array of 48 small cameras. Each camera images a portion of the scene or sample. The multi-camera configuration works like a single microscope, capturing high-resolution, gigapixel images of large and non-flat objects in a single snapshot. Each camera can be independently focused to match the sample surface, ensuring that the entire field of view stays sharp even if the sample is curved. Additionally, the multi-camera, curvature-adaptive microscope also eliminates the need for scanning, which can take up to an hour. In a process that takes about 5-10 min, PANORAMA uses software to automatically stitch the images from each camera together into one continuous picture. “The telecentric lens makes it possible to image a very wide field without distortion, while the multi-camera approach overcomes the usual size-and-resolution limit of a single sensor,” Chen said. “This combination lets us acquire a seamless, gigapixel image in a single snapshot, flattening out any curvature adaptively.” The detailed, gigapixel-scale images have 10-50x more pixels than the average smartphone camera image. Imaging a prepared slide of rat brain tissue under brightfield illumination, which uses white light to reveal tissue structure, enabled the researchers to demonstrate the efficacy of the instrument and technique. The 48-camera array captured the entire 630-megapixel (MP) image in one snapshot, with no scanning required. The resulting image showed cellular structures as small as 0.84?µm, as well as neurons and dendrites across the sample. The microscope combines a telecentric lens, a tube lens, and an array of micro-cameras, each with its own focus control. This configuration makes it possible for the microscope to acquire a seamless, gigapixel image in a single snapshot, flattening out any curvature adaptively. Courtesy of Duke University/Haitao Chen. The researchers also used PANORAMA to simultaneously acquire brightfield and fluorescence images of onion skin placed over a curved surface. By focusing each camera on the local curvature, they were able to obtain sharp images of the entire onion skin over the curved surface. The brightfield images revealed crisp cell walls, while the fluorescence images clearly showed stained nuclei. “In practical terms, we saw a huge jump in throughput and flexibility — no more moving parts, no tedious focus-stacking, and no blind spots between cameras,” professor Roarke Horstmeyer, who led the research, said. “Compared to older multi-camera microscopes that needed scanning to fill gaps and maintain focus, our approach gives continuous full coverage at sub-micron resolution.” The researchers are investigating how to improve the microscope by adding more cameras or larger sensors to capture an even bigger field, such as an entire petri dish, in a single shot. They are also developing an automated focus system, which will eliminate the need to adjust each camera manually for every sample. Future computational advances could make it possible for PANORAMA to perform 3D image reconstruction, provide depth maps in real time, and provide live videos of microscopic processes. “Although traditional microscopes assume the sample is perfectly flat, real-life samples such as tissue sections, plant samples, or flexible materials may be curved, tilted, or uneven,” Horstmeyer said. “With our approach, it’s possible to adjust the focus across the sample, so that everything remains in focus even if the sample surface isn’t flat, while avoiding slow scanning or expensive special lenses.” The research was published in Optics Letters (www.doi.org/10.1364/OL.572466).