Search

Stroke is commonly caused by thromboembolic events originating from atherosclerotic plaques in the carotid vasculature. To improve stroke risk prediction by delineating vulnerable plaque features, our group has developed ARFI Variance of Acceleration log(VoA) imaging. We herein evaluate the ability of trained, blinded readers to interpret in vivo ARFI log(VoA) images of human carotid plaques, with validation by matched histology. Patients undergoing carotid endarterectomy were imaged using a Siemens S3000 and a 9L4 transducer. From 21 plaques, one was excluded due to specimen damage during surgery. Parametric 2D images of ARFI log(VoA) were rendered and evaluated by 3 neuroradiologists, 1 abdominal radiologist, 1 sonographer, and 1 pathologist, all blinded to the histological gold standard. Receiver operating characteristic (ROC) curve analysis was performed, and area under the curve (AUC) was taken as a metric of performance for detecting lipid-rich necrotic core (LRNC), intraplaque hemorrhage (IPH), collagen (COL), and calcium (CAL). Reader log(VoA) outcomes were compared to those achieved using ARFI peak displacement (PD) in a comparable prior study. Average AUC performance for plaque features were as follows: CAL, 0.77; COL, 0.76; LRNC, 0.79; IPH, 0.82. Grouping the stiff (COL and CAL) and soft (LRNC and IPH) features together increased average AUCs to 0.86 for stiff and 0.87 for soft. These log(VoA) AUC outcomes were higher than those achieved by blinded readers evaluating ARFI PD images. This study shows the relevance of ARFI log(VoA) imaging, as interpreted by trained, blinded readers, to delineating the structure and composition of carotid atherosclerotic plaque in humans, in vivo.

Rupture potential of atherosclerotic plaques in carotid arteries is conferred by both composition and structure of plaques. Previous studies have shown that from in vivo collected data, carotid plaque components such as collagen, calcium, necrotic core and intraplaque hemorrhage can be automatically detected by an ARFI imaging-derived machine learning classifier. Automatic classification considered normalized cross-correlation measurements of ARFI-induced displacement, signal-to-noise ratio and cross-correlation coefficients from an on-axis ARFI acquisition. We now extend our prior work by hypothesizing that using multiple ARFI data forms improves plaque feature detection and FC thickness measurement in human carotid plaques, relative to a single form of ARFI data. Carotid plaques were imaged in vivo prior to surgery in 20 patients undergoing carotid endarterectomy (CEA), and extracted plaque specimens were harvested after CEA for histological processing. ARFI data were acquired with cardiac gating to diastole with push (DP) and to systole without push (SNP) using fundamental low (FL), fundamental high (FH), and harmonic (H) tracking frequencies. Combinations of the resulting displacement profiles were used as inputs to the SVM classifier. The classifier was evaluated by 5-fold cross-validation, with the histological samples acting as gold standards. From the output SVM likelihood matrices, ROC curves were calculated for separating collagen from calcium and lipid-rich necrotic core from intraplaque hemorrhage. For all examined plaques, DP+FH+H achieved the highest average AUC of 0.926 (Sensitivity = 0.932, Specificity = 0.915) but required 2 acquisitions. The best data combination requiring only one acquisition (DP+FH) achieved an AUC of 0.917 (Sensitivity = 0.902, Specificity = 0.880). These results suggest that using multiple forms of frequency and gating of ARFI data as inputs to an automatic classifier improves discrimination of carotid plaque components, relevant for rupture vulnerability assessment.

Quantitative Viscoelastic Response (QVisR) ultrasound uses in silico on-axis VisR displacement profiles to estimate the elastic and viscous moduli of an interrogated material. This work studies the effect of varying the radio frequency (RF) data signal-to-noise ratio (SNR) on QVisR's estimates. White Gaussian noise was added to the RF data of simulated viscoelastic materials subject to a VisR beamsequence with SNRs from 0-50dB in steps of 10dB. Tracked displacements were used as features for a neural network model to estimate elastic and viscous moduli. Predicted elastic moduli RMSE was smaller than the simulated material elastic step size down to 20dB SNR and viscous moduli RMSE was smaller than the viscous step size down to 40dB SNR. These findings suggest QVisR can be used to estimate viscoelastic moduli in noisy environments that more closely model non-simulation data acquisitions.

Acoustic Radiation Force Impulse (ARFI) Log Variance of Acceleration (log(VoA)) was previously demonstrated with a single ARFI excitation and plane wave (PW) tracking to detect plaque structure and composition. However, a separate multi-angle sequence was required to enable simultaneous display of 2-D vector-Doppler (VD) and plaque morphology. This paper evaluates the performance of various multi-angle compounded PW implementations of log(VOA). Experiments were conducted in a CIRS elastic phantom and in an excised human cadaveric carotid. In phantoms, increasing SNR by way of compounding increased inclusion SNR and CNR. Increasing the angle range over which PWs were steered decreased SNR, but did not appreciably change CNR. When applied to the ex vivo carotid, an angle range of 10° best predicted calcium area (as validated by spatially-aligned, expert-segmented histology), and higher number of angles spanning 10° improved delineation of plaque boundaries. These results suggest that, compared to an unsteered PW sequence, a multi-angle compounded PW sequence improves log(VoA) performance, suggesting the potential for extension to implementation with concurrent vector Doppler imaging.

Double-profile intersection (DoPIo) ultrasound combines two displacement profiles capturing identical tissue motion following an acoustic radiation force (ARF) push to estimate shear elastic modulus via an empirically derived model. However, the displacement-tracking scheme may be impacted by differences in focal configurations for both the push and track beams. A wider push beam imparts a more uniform displacement gradient than narrow ARF pushes for on-axis tracking beams, while the simultaneous formation of two displacement profiles from a single, wide transmit pulse enables the tracking of identical scatterer distributions at the cost of diminished differences between the two displacement profiles. In silico experiments suggested that DoPIo acquisitions performed using a wide ARF push beam and simultaneous tracking provided the most accurate and precise elasticity estimates. While all four combinations of parameters enabled the differentiation of a soft inclusion within a stiff background in vitro, elasticity estimates on a commercially calibrated phantom were consistently overestimated.