Deep learning-enhanced autofluorescence-harmonic imaging. (IMAGE)
Caption
a A 5.4 × 5.4 mm² area image of human ovarian borderline carcinoma merged by 3PA of NADH (blue), SHG of collagen (green), and 2PA of FAD (red). The white dashed line separates the registered input image and network output image. White squares indicate ROIs magnified in b–d, which compare the registered input (left) and network output (right). White dashed arrows in (b) indicate the SFA. White ellipses indicate the blood vessel. e Intensity profiles are along the white solid line in (a). Black dashed lines separate the signal (S) and noise (N) regions according to the ROI above. f SNR of the raw input and network output images obtained using the ground truth as ideal images. A two-tailed Wilcoxon matched-pairs signed rank test was applied between the input and output images. n = 12 test large images. g Acquisition and inference time of DLAM compared to GT acquisition time. Scale bars, 500 μm in a and 200 μm in b–d.
Credit
by Binglin Shen, Shaowen Liu, Yanping Li, Ying Pan, Yuan Lu, Rui Hu, Junle Qu, Liwei Liu
Usage Restrictions
Credit must be given to the creator.
License
CC BY