FIGURE 5 (a) The panel shows US image and low no. of frame average PA images of a sample in-vitro graphite rod phantom (left panel), and a subcutaneous mouse tumor (right panel), respectively. The background is indicated by the orange square and the signal ROI is denoted by the green square. (b-d) The box plot shows the SNR, PSNR and CNR values over the specified ROI in the image (a) for the low number of frame averaged LED acquired images and the images that are predicted by the U-Net deep learning technique. There is significant improvement (p<0.01) in the SNR, PSNR and CNR compared to the corresponding low number of frame averaged images. (e) A representative low number of frames averaged LED image of a mouse subcutaneous tumor. (f-h) The corresponding SNR, PSNR and CNR plots, respectively for the images acquired on subcutaneous tumors. Data points beyond the interquartile range are displayed using red plus. Student’s t-test has been used for the statistical analysis of all the data.