An open-source GPU-based real-time quantitative ultrasound toolbox

Quantitative ultrasound (QUS) is an imaging technique, which includes spectral-based parameterization and envelope statistics. In terms of spectral-based QUS, the backscatter coefficient (BSC) and attenuation coefficient slope (ACS) are frequently utilized because they can be systemindependent parameters and are fundamental properties of tissues. Many previous studies proposed to make QUS imaging more accurate, by using regularization or full angular spatial compounding. However, few studies have focused on optimizing the processing time, which is important in clinical scenarios. The need to optimize the processing time is more demanding when dealing with three dimensional volumetric QUS imaging, or with multiple frames. Because the spectral-based QUS relies on estimating a large number of Fourier transforms (FFTs) for each data block, and modeling the spectra using physical assumptions and approximations, the large numbers of FFTs can be calculated in parallel with a GPU, and the further calculations can also be optimized to improve processing speed. As a result, the processing of QUS images can be faster and approach real time with a GPU. In this study, an open-source GPU-based real-time quantitative ultrasound toolbox was developed, which is intended not only for researchers, but also clinical use. The code and software are available publicly on: https://github.com/Ultrasound-Research-Laboratory/Open-GPU-QUS