We here report an optical system for live 3D microscopy that we call the Multifocus 25-camera microscope (M25). M25 is a new design for aberration-corrected multifocus microscopy (MFM) that employs the latest generation of small, fast and sensitive CMOS cameras. Each of the 25 focal planes is captured on one of the individual cameras, enabling truly simultaneous 3D recording at >100Hz. M25 is built from a combination of custom manufactured diffractive Fourier optics elements and off-the-shelf components. We are employing M25 to study neural circuit function in small model organisms including C. elegans, Drosophila, and fish.
Microscopic study of rapid biological processes often requires both high resolution and high acquisition speed. When the speed requirement precludes acquiring a full 3D focal series at each time point, it can be attractive to sacrifice all axial information and instead record a single, 2D image per time point. This can be done at very high frame rates. High-resolution objectives, however, have a very short depth of focus. There are several established methods to achieve extended depth of focus, including annular pupil masks; mechanical sweeping of the focus and wavefront coding, which uses a pupil-plane optical device to introduce geometric aberrations. We have developed a new pupil plane approach where the light is manipulated chromatically rather than geometrically. A phase mask with circularly symmetric stair steps divides the pupil plane into a series of annular zones. The stair steps are large compared to the coherence length of the observation light, so that images from different zones form independently and combine incoherently into a final image. Each zone carries only a fraction of the objective's axial resolution, but the larger zones still carry the full lateral resolution of the objective. The incoherent addition of the different single-zone images results in a smooth and circularly symmetric point spread function with a depth of focus that is extended by a factor approximately equal to the number of zones in the mask. The method has been demonstrated both on bead samples and on whole cells with a performance that is well in accordance with the theoretical predictions.