Skip to main content
. 2019 Nov 14;9:16770. doi: 10.1038/s41598-019-53217-y

Figure 4.

Figure 4

Examples of images from videos of semen samples with different concentrations (columns) and the four image representations used to train the neural network-based algorithms (rows). Sperm concentration; (A) 4 per x106/mL, (B) 33 per x106/mL, (C) 105 per x106/mL, (D) 192 per x106/mL, and (E) 350 per x106/mL. Image representation; (1) original video, (2) sparse optical flow, (3) dense optical flow, and (4) vertical frame matrix.