I'm working on a project to improve our current PAT system and image-processing routine in particular, and I've hit a wall.
We use a 6mm diameter, unfocused, 10 MHz transducer in a raster-scanning configuration in an imaging plane above the object. The issue I'm having is implementing the reconstruction after the data has been collected. In principle, if the imaging surface is a grid with NxM points, then image formation will involve NM separate reconstructions, all of which will then need to be combined to form a single image. And this is assuming that I can treat the initial pressure distribution as uniform throughout the data collection process (this assumption seems reasonable - the irradiation area is large relative to the size of the objects we typically study - but I'm not 100% convinced it's accurate). Then I will need to correct for imaging artifacts that result from the NxM separate limited FOV measurements. All of this seems prohibitively expensive computationally, and severely limited in terms of the accuracy and resolution we could hope to achieve even in the best case scenario.
Has anybody implemented accurate single-element raster-scanning PAT imaging systems? I know the setup is common in PAM, but tomography introduces its own set of challenges.
Thanks,
Brendan