Ultrasound-guided identification of cardiac imaging windows
Diagram of sample data from our multimodal experiment. Top left: Four-chamber apical view TDE of basal IVS. Top middle and right: Corresponding orthogonal biplane angiographic images of an RCA. The circular object in the top right frame is the photo-x-ray sensitive PIN diode used to record the timing of x-ray pulses. Bottom: Timing plot of postprocessed velocity data for one heartbeat. The E and A peaks are labeled on the IVS velocity plot (thick black line) indicating the early and late ventricular filling phases, respectively. The post-processed 3D velocity of the coronary artery is shown in three plots (thin lines) corresponding to the proximal, mid, and distal segments. The proximal and distal segments have the highest and lowest A-peak velocities, respectively. This is perhaps reflective of each segment’s proximity to the atrium during atrial contraction. In this example, the estimated vessel blur during the IVS gating window was 0.79 mm.
Modified Bland–Altman plots showing good agreement between IVS and x-ray gating windows. Each point represents a single patient in the study. Error bars indicate variability across heartbeats. On average, the start of IVS gating windows lag that of x-ray gating windows by 12 ms; the end of IVS gating windows lag that of x-ray gating windows by 3.5 ms.
Plots showing very high correlation between the timing of IVS versus x-ray gating windows. Both plots have R > 0.99 with p < 0.01.
Correlation plots between the timing parameters of x-ray diastasis gating windows versus the corresponding RR intervals. Left: R = 0.63 (p > 0.1). Right: R = 0.86 (p < 0.02). These plots show that heartbeat duration is an imprecise predictor of when coronary quiescence occurs in diastole.
Frontal (left) and lateral (right) plane PSFs of the distal ROI during IVS gating windows across two heartbeats in patient #3. Each point on the PSF represents the fraction of time the ROI spends at a location. The dotted line is a contour that surrounds 70% of the motion blur. The average diameters of the 70% energy contour are 0.31 and 0.44 mm for the frontal and lateral planes, respectively. Adding these numbers in quadrature yield the combined estimate of an ROI blur of 0.54 mm.
Blur agreement between x-ray and IVS gating windows. The modified Bland–Altman plots show good agreement between coronary artery blurring caused by x-ray and IVS gating windows over single (left) and multiple (right) heartbeats. Each point represents data from a single patient. These plots show that the two gating methods can be expected to render images of comparable spatial resolution.
Start times, end times, and durations of IVS and x-ray gating windows relative to the R peak on the ECG.
Summary of vessel blur during IVS and x-ray gated windows across single and multiple heartbeats. Vessel blur estimates are reported as an average between proximal, mid, and distal ROI blur estimates. Overall, vessel blur was highly correlated between the IVS and x-ray gating techniques: R = 0.99 (p < 0.01) for the single heartbeat case and R = 0.91 (p = 0.03) for the multiple heartbeat case.
Summary of MAVDvessel and MPLDvessel, as determined by the x-ray images, and MAVDivs and MPLDivs, as determined by the TDE data. The above measurements were made from the first heartbeat of each patient.
Summary of correlation coefficients between vessel blur and the set of vessel blur predictors. For both the IVS and x-ray gating techniques, vessel blur is not correlated with heartbeat duration, nor MAVD, but is weakly correlated with diastasis duration and strongly correlated with MPLD. This result highlights the potential use of monitoring MPLDivs in an IVS gating system to predict the sharpness of the image data acquired during the IVS gating window. Certain combinations for correlation such as MPLDivs versus vessel blur during the x-ray gating window are omitted in this analysis because they are not clinically meaningful.
Article metrics loading...
Full text loading...