Deep learning-based digital refocusing in acoustic microscope
Scanning acoustic microscopy (SAM) is a label free imaging technique which is capable of visualizing the surface and sub-surface structure from materials to biological samples. In acoustic imaging, synthetic aperture focusing technique (SAFT) that was developed in the 1970’s is one of the most versatile techniques to solve the focusing issue. The optical holography and synthetic aperture radar technologies served as inspiration for the ultrasonic SAFT reconstruction. There are some short coming in SAFT for reconstructing the SAM images- like it generates line artifact in the reconstructed images. It also requires defocusing distance of the sample to focal point. To overcome such issue, we proposed a physics based deep learning (DL) approach to effectively refocus the images. Point spread function (PSF) has been obtained from COMSOL Multiphysics, which was used to defocus the images. In this project we are proposing the out of focus images can be corrected by employing DL.