Low Velocity Blood Flow Imaging using Aperture Domain Clutter Suppression Techniques

Abstract

Transarterial Chemoembolization (TACE) is a locoregional therapy for liver cancer that reduces blood flow within tumor-feeding vessels using an embolic agent. Assessment of therapeutic efficacy is performed four weeks post-treatment using contrast-enhanced MRI or CT to detect the change in tumor contrast uptake. Earlier assessment is unreliable due to confounding treatment-induced artifacts; this results in delayed retreatment and can lead to interim disease progression. Non-contrast ultrasound imaging can detect changes in blood flow immediately after TACE and may be a preferential modality for early assessment of therapeutic outcomes. Recent advancements in power Doppler ultrasound, including plane wave sequences and motion correction, have enabled detection of low velocity blood flow. Nonetheless, clinical imaging remains difficult due to insufficient removal of thermal noise and acoustic clutter. Power Doppler conventionally relies on temporal filtering of the echo data to isolate the blood signal, but filter utility suffers when blood, clutter, and noise exhibit similar frequency domain characteristics. However, these signals exhibit different characteristics in the aperture domain and can be separated by comparing the echoes recorded at different elements of the transducer. To improve low velocity blood flow imaging, we propose two technologies that leverage aperture domain characteristics. First, we adapted the Coherent Flow Power Doppler (CFPD) image formation technique to preserve the relationship between the image intensity and blood volume, which permits sensitivity toward TACE-induced changes in blood flow. Our technique, called power-preserving CFPD employs a measure of spatial covariance to reduce the contribution of noise and clutter. Second, we developed an adaptive filter that employs a higher-order singular value decomposition (HOSVD) applied to the tensor of aperture data. We present temporal, spatial, and aperture domain features that can be leveraged in filtering and demonstrate that this multidimensional approach improves sensitivity toward blood flow. These technologies were successfully validated using simulation, phantom, and in vivo data. Finally, these methods were combined with other advanced Doppler methods for a preliminary study of patients undergoing TACE. We demonstrate that non-contrast power Doppler imaging has the potential to detect residual vascularity and differentiate therapeutic outcomes after TACE.