Assessing the performance of vessel wall tracking algorithms: the importance of the test phantom

There is widespread clinical interest in assessing the mechanical properties of tissues and vessel walls. This study investigated the importance of the test phantom in providing a realistic assessment of clinical wall tracking performance for a variety of ultrasound modalities. B-mode, colour Doppler and Tissue Doppler Imaging (TDI) cineloop images were acquired using a Philips HDI5000 scanner and L12-5 probe. In-vivo longitudinal sections of 30 common carotid arteries and in-vitro images of pulsatile flow of a blood mimicking fluid through walled and wall-less tissue and vessel mimicking flow phantoms were analysed. Vessel wall tracking performance was assessed for our new probabilistic B-mode algorithm (PROBAL), and 3 different techniques implemented by Philips Medical Systems, based on B-mode edge detection (LDOT), colour Doppler (CVIQ) and TDI (TDIAWM). Precision (standard deviation/mean) of the peak systole dilations for respective PROBAL, LDOT, CVIQ and TDIAWM techniques were: 15.4 ± 8.4%, 23 ± 12.7%, 10 ± 10% and 10.3 ± 8.1% for the common carotid arteries; 6.4%, 22%, 11.6% and 34.5% for the wall-less flow phantom, 5.3%, 9.8%, 23.4% and 2.7% for the C-flex walled phantom and 3.9%, 2.6%, 1% and 3.2% for the latex walled phantom. The test phantom design and construction had a significant effect on the measurement of wall tracking performance.