Deep Learning-Based Virtual Phase Contrast Imaging Method

Fig. 1. Schematic of conventional Zernike phase contrast imaging principle (1--4 represent front focal plane, phase plate, and main phase plane, respectively)

Download full size

Fig. 2. Schematic of GANs model

Download full size

Fig. 3. Schematics of the Cycle-GANs model^[16]. (a) The model contains two mapping functions (G:X→Y and F:Y→X) and two adversarial discriminators (D_Y and D_X); (b) forward cycle-consistency loss; (c) backward cycle-consistency loss

Download full size

Fig. 4. Flow chart of virtual phase contrast imaging

Download full size

Fig. 5. Virtual phase contrast imaging results. (a)--(c) Bright field images acquired by microscope; (d)--(f) virtual phase contrast images output by neural network calculation; (g)--(i) phase contrast images corresponding to bright field images acquired by microscope

Download full size

Table 1. Objective evaluation indicators^[26]
View table
Table 1. Objective evaluation indicators^[26]
Evaluation indicator Formulaic expression
COSIN $\frac{\overset{n}{\sum_{i = 1}} A_{i} \times B_{i}}{\sqrt[]{\overset{n}{\sum_{i = 1}} {(A_{i})}^{2}} \times \sqrt[]{\overset{n}{\sum_{i = 1}} {(B_{i})}^{2}}}$
PSNR $\frac{255^{2}}{\overset{M}{\sum_{i = 1}} \overset{N}{\sum_{j = 1}} {[F (i, j) - R (i, j)]}^{2}}$

Table 2. Objective evaluation of virtual phase contrast imaging and standard phase contrast imaging results
View table
Table 2. Objective evaluation of virtual phase contrast imaging and standard phase contrast imaging results
Image group Objective evaluation indicator
COSIN PSNR
1 0.9216 40.1234
2 0.9269 39.9938
3 0.9039 40.1857

Tools

Get Citation

Copy Citation Text

Zhongfa Liu, Yizhe Yang, Yu Fang, Xiaojing Wu, Siwei Zhu, Yong Yang. Deep Learning-Based Virtual Phase Contrast Imaging Method[J]. Acta Optica Sinica, 2021, 41(22): 2217001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites