DeepH, a new method for fast harmonic ultrasound imaging based on Deep Learning , is published by the IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control and made on the front cover of the March issue.

Our article presents a new processing method to achieve high quality clinical harmonic ultrasound images which faster acquisition schemes.

Harmonic Imaging is the current standard imaging modality for many clinical applications of ultrasound imaging. The nonlinear sound propagation in tissues generates harmonic frequencies of the originally transmitted sound pulse. This allows to record images with better spatial resolution and higher contrast. However, a sequence of several acquisitions is needed to generate a harmonic image. This lowers the framerate and makes images sensitive to patient motion.

In our article we demonstrate that the DeepH deep neural network can process one acquisition of an ultrasound image to generate a harmonic image from it. The image detail is comparable to the ground truth harmonic image. Additionally, noise is suppressed and the framerate is increased.

Find the full article on DeepH for fast harmonic imaging published at DOI: 10.1109/TUFFC.2023.3234230

Fouad, M., Abd-El-Ghany, M., Schmitz G.: “A Single-Shot Harmonic Imaging Approach Utilizing Deep Learning for Medical Ultrasound”