Deep Learning for Image Processing and Reconstruction to Enhance LED-Based Photoacoustic Imaging

Abstract

Photoacoustic imaging is a rapidly growing imaging technique which combines the best of optical and ultrasound imaging. For the clinical translation of photoacoustic imaging, a lot of steps are being taken and different parameters are being continuously improved. Improvement in image reconstruction, denoising and improvement of resolution are important especially for photoacoustic images obtained from low energy lasers like pulsed laser diodes and light emitting diodes. Machine learning and artificial intelligence can help in the process significantly. Particularly deep learning based models using convolutional neural networks can aid in the image improvement in a very short duration. In this chapter we will be discussing the basics of neural networks and how they can be used for improving photoacoustic imaging. We will also discuss few examples of deep learning networks put to use for image reconstruction, image denoising, and improving image resolution in photoacoustic imaging. We will also discuss further the possibilities with deep learning in the photoacoustic imaging arena.