Session Index

Nowadays, hyperspectral imaging systems are costly and difficult to produce. Besides, the measurement is extremely time-consuming, which is not conducive to the rapidly changing scientific measurement. In this paper, we proposed a hyperspectral imaging system with reasonable production cost and fast measurement speed

 

A series of photopolymers based on the ethylene glycol phenyl ethyl arylate (EGPEA) monomers with varying initiator concentration and sample thickness is synthesized. The advantages of lowering the initiator concentration, including a rather short initiation time within few seconds and a sharp rising optical response, are demonstrated.

 

Airy beam light-sheet microscopy offer advantages in providing better optical sectioning capability, longer field of view, and deeper penetration depth. With the deconvolution method, contrast and resolution in images can be improved by removing out-of-focus noise. Here, we demonstrated Airy light-sheet microscopy and used the Richardson-Lucy deconvolution method to further improve the image quality.

 

In this paper, we present the feasibility of computer generated hologram (CGH) by Neural Network. In our method, CGH is done by autoencoder framework and the training dataset of the deep learning model is created by Iterative Fourier Transform Algorithm (IFTA). The method enables the non-iterative calculation of CGH for real-time applications.

 

Multiphoton excitation can reduce tissue scattering and get deeper image depth. With optogenetics, neural activity can be modulated to influence biological behavior. Precisely spatial and temporal control the shape of lights is important to observe the 3D distributed neural connection. In this study, we use a deep-learning based computer-generated holography to generate holograms with orders of magnitude faster and better accuracy. Furthermore, to improve axial resolution of the stimulated pattern, we introduce temporal focusing (TF) to get a better axial confinement. With all the applied technology, we can real-time and precisely stimulate drosophila brain in single-cell resolution.

 

We used generative adversarial networks (GAN) to help phase retrieval from interferometry. The established image is transformed from fringe pattern to phase map. They can be fitted by Zernike polynomial and reducing the calculating steps.