High Frequency Ultrasonic Tomography Using Optimal Transport Distance

Nonlinear tomographic ultrasound reconstructions can provide quantitative data about material parameters like the speed of sound also in reflection-mode imaging. However, these methods suffer from the cycle skipping issue which is caused by the high center frequencies usual in medical ultrasound hardware. As these algorithms are based on minimization of a misfit function where the high frequency signals lead to local minima, false reconstruction are the result. Previous approaches try to circumvent this problem by low pass filtering the measured signals. This method is limited by the bandwidth and SNR of the measurement system. In this contribution, we applied the optimal transport distance known from seismic reconstructions. This approach uses all frequency components while enhancing the included traveltime information. We compare the misfit functions and show reconstructions using the standard and the optimal transport method. The results demonstrate an at least 1.6 times higher convergence limit.