Temporal-focusing two-photon microscopy is a wide-field fluorescence imaging technique with depth sectioning capability. We developed 3D imaging without laser beam scanning or mechanical motion by integrating pulse shaping together with temporal focusing. This microscope is capable of high-speed volumetric imaging controlled by the electronic signals generated from an arbitrary function generator. The fast 3D imaging capability of the microscope is applicable to biomedical studies such as tracing cellular dynamics in live animals, where conventional scanning microscopes are limited by slow mechanic scanning.
Two-photonflow cytometer was developed a decade ago. Its sampling volume is small due to the fact that efficient two-photon excitation happens at the focal point of a focused Gaussian beam, which makes it challenging to count rare circulating cells. Airy beams have comparable beam spot size with Gaussian beams, and can propagate for a long distance without diffraction. Hence, scanned Airy beams can form a large light sheet to drastically increase the sampling volume. We developed a two-photon flow cytometer based on 2D Airy beam light-sheet and demonstrated its capability of detect circulating beads and cells in microfluidic channels.
Object detection and semantic segmentation are classical problems in computer vision. In recent years there are great successes of implementing convolutional neural networks (CNNs) in this field. We have developed a fast and fully-automated end-to-end solution that can efficiently segment lung nodule contours from raw thoracic CT scans. Furthermore, this novel CNN architecture has be applied on super-resolution image reconstruction for photoactivated localization microscopy (PALM) and stochastic optical reconstruction microscopy (STORM).