Deep Learning-Based Methods for Photoacoustic Imaging Reconstruction: Concepts, Promises, Pitfalls, and Futures

The methods currently available for performing deep learning rely a lot on developing models that are reproducible as well as getting the required amount of data for performing deep learning. Here, we have done a review keeping in mind the fact that the data requirement and the model availability and how the computer vision models have translated to the medical imaging scenarios in the current world. This chapter consists of dividing the whole domain into the reconstruction part and then looking forward to how the models from the natural images adapt to the photoacoustic imaging in terms of changing the architecture, working on modifying the architectures as well as looking for the required dataset for performing deep learning on the photoacoustic image reconstruction. We discuss more on the architecture sides of deep learning as well as what are the future directions where deep learning can be utilized for improving photoacoustic imaging.