To further the use of programmable illumination for computational microscopy methods, researchers at the University of Connecticut constructed, calibrated, and tested freeform illuminators in different forms — in the shapes of a tilted flat surface, a triangular pyramid, a dome, and a Mobius band. The freeform illuminators, with the calibration process, offer flexibilities and extended scope for imaging innovations in computational microscopy, as well as greater efficiency. “With this platform, we can start to transit Petri-dish-based experiments from the traditional labor-intensive process to an automated and streamlined process,” the researchers said. Programmable light sources simplify the work of illuminating samples in different kinds of microscopy contexts, but conventional programmable arrays consist of a flat, fixed grid of individual lights. An array consisting of lights that can be placed anywhere in three dimensions offers opportunity to reduce the size of the illuminator, to place the lights closer to the sample, and/or to increase the density of the lights and adjust the angle of illumination according to users’ requirements. The team developed a calibration process using a low-cost image sensor coated with a layer of blood cells. The blood-cell monolayer on the sensor coverglass modulates the incoming lightwaves and generates a diffraction pattern on the detector plane. The researchers inferred the 3D positions of the light source elements by tracking the positional shift of the blood-cell diffraction patterns at two distinct regions in the coded sensor. This approach is similar to the way that stereo vision reconstruction works. The researchers tested the calibration method and found that the recovered positions matched the actual positions with only small deviations. Human blood from a finger prick was used to coat the sensors — no expensive, sophisticated tools were required. Substances with similar properties could be used in place of blood, according to the researchers. Once calibrated, the freeform illuminators were ready to test with different computational microscopy techniques. The researchers demonstrated the illuminator with Fourier ptychographic microscopy (FPM), 3D tomographic imaging, and on-chip microscopy. They also used the illuminator in an experiment to track bacterial growth over a large field of view. All four illuminators were used successfully for FPM, a computational microscopy technique that combines different images produced with lighting at varying angles. The pyramid illuminator was used to capture separate images lit with red, green, and blue light, which were combined to make a full-color image. The team said that combining the images produced with separate lights resulted in a final image with higher resolution than any of the raw images alone. Tests showed that the pyramid design, which was angled toward the sample, held the highest number of lights at the point farthest from the sample. Due to its small footprint, the pyramid illuminator could be placed close to the sample to deliver light efficiently. The pyramid illuminator was also used for 3D tomography, a technique that builds a 3D image by stacking cross-section images of a 3D sample. The researchers also tested the pyramid design with superresolution, on-chip microscopy, a technique in which the sample sits directly on the sensor. In a longitudinal study, the researchers produced a time-lapse series of the growth of a colony of E. coli bacteria. Experimental setups using the freeform illuminator could bring more flexibility and greater efficiency to computational microscopy. In the future, the researchers said, freeform illuminators could be enhanced by improving calibration methods, increasing the density of lighting arrangements and the diversity of wavelengths, and adjusting position estimates. The research was published in Intelligent Computing (www.doi.org/10.34133/icomputing.0015).
