Volumetric blood flow imaging with ultrafast ultrasound: a new approach based on the speckle decorrelation method

The estimation of 3-D blood flow through vessels represents one of the most challenging aspects both in clinics and in research. Through the measurement of the volume flow it is possible to prevent many cardiovascular diseases, which are among the main causes of deaths all over the world. Currently, because of the limitations existing in the modern ultrasound systems, there is no efficient solution to estimate the volume flow. In recent years an interesting approach, capable to evaluate the volumetric blood flow from the cross-sectional imaging plane was conceived. This approach is based on detecting the out-of-plane velocity component by using the speckle decorrelation (SDC) and the in-plane velocity components by using the ultrasound imaging velocimetry (UIV). By making use of micro-bubbles contrast agents and a standard 1-D array transducer working at high frame-rate, the recent research has found compelling results. Despite the good results, the new technique performs well only in presence of micro-bubbles contrast agents, which are necessary to enhance the blood flow signal. Therefore, this master thesis project aims to test the goodness of the speckle decorrelation method in absence of micro-bubbles. Since blood is anechoic, meaning that it produces weak echoes when hit by ultrasound, it is necessary to find a way to increase its signal. Singular value decomposition (SVD) filtering has been performed to achieve this result. The proposed method has been evaluated on wall-less flow phantom, with the help of a pulsatile pump, which emulates the heartbeat. The phantom size is close to the one of big human vessels. In vitro experiments have been conducted with the use of the blood mimicking fluid (BMF), which is able to mimic the characteristics of human blood. The influence of different combinations of voltage and pulse length on the estimation of the velocity has been studied, as well. The conducted experiments show interesting results. In particular, it has been proved that at a flow rate of 80 mL/min and with low values of pulse length, i.e. PL=1 and PL=2, the estimated volume flow matches well with the ground-truth, providing an averaged error of 3.90% ± 2.43%. On the other side, the results obtained with high value of pulse length, i.e. PL=4 and PL=6, provided an averaged error of 16.01% ± 7.91%. The in vitro trials suggest that the technique studied in this master thesis project could give an important contribution in the next future clinical applications.