Fig. 4

The results of quantitative assessment for image enhancement. A and B are the relationships (means and covariances in 205 images) between quantitative metrics and number of frames used in traditional averaging (blue lines) and deep learning method (red lines). C is CNR comparisons (means and covariances in 205 images) between image averaging (blue bar, Avg-5) and deep learning method (red bar, DL-5). D is subjective scoring (means values in 205 images) for image quality enhanced by traditional averaging (blue line, Avg-5) and deep learning method (red line, DL-5) for normal retinal structure and retinal lesions