Simulating compressive three-dimensional structured illumination microscopy

Abstract

Three-dimensional structured illumination microscopy (3D-SIM) requires at least fifteen raw images to extend the spatial resolution to the half of the diffraction limit in both axial and lateral directions. This requirement can cause motion artifacts in a reconstructed super-resolution image. Therefore, reducing the acquisition duration of raw images is of interest. Here, we propose a novel method, which applies compressed sensing (CS) to 3D-SIM and enables faster raw image acquisitions. CS allows us to record a scene with a photomultiplier tube, which is much faster than a camera. Furthermore CS can recover the scene from a small number of measurements. The combination of these two facts leads to significant reduction of the overall acquisition time. In preliminary experiments, a computer-generated ground truth image is utilized to simulate the proposed method, and the spatial resolution of an image produced by the proposed method is compared with conventional optical microscopy and 3D-SIM.