Blood velocity measurements are important in physiological science and clinical diagnosis. Doppler ultrasound is the most commonly used method but can only measure one velocity component. Ultrasound imaging velocimetry (UIV) is a promising technique capable of measuring two velocity components; however, there is a limit on the maximum velocity that can be measured with conventional hardware which results from the way images are acquired by sweeping the ultrasound beam across the field of view. Interleaved UIV is an extension of UIV in which two image frames are acquired concurrently, allowing the effective interframe separation time to be reduced and therefore increasing the maximum velocity that can be measured. The sweeping of the ultrasound beam across the image results in a systematic error which must be corrected: in this work, we derived and implemented a new velocity correction method which accounts for acceleration of the scatterers. We then, for the first time, assessed the performance of interleaved UIV for measuring pulsatile arterial velocities by measuring flows in phantoms and in vivo and comparing the results with spectral Doppler ultrasound and transit-time flow probe data. The velocity and flow rate in the phantom agreed within 5–10% of peak velocity, and 2–9% of peak flow, respectively, and in vivo the velocity difference was 9% of peak velocity. The maximum velocity measured was 1.8 m s−1, the highest velocity reported with UIV. This will allow flows in diseased arteries to be investigated and so has the potential to increase diagnostic accuracy and enable new vascular research.@en

It is not practical to obtain a large number of labeled data to train a supervised learning network in tunnel lining nondestructive testing with ground-penetrating radar (GPR). To decrease the dependence of supervised learning on the number of labeled data, an improved self-supervised learning algorithm—self-attention dense contrastive learning (SA-DenseCL)—is proposed and incorporated with a mask region-convolution neural network (Mask R-CNN), which is trained by unlabeled and labeled GPR data. The proposed SA-DenseCL adds a self-attention-based relevant projection head to the DenseCL architecture of self-supervised learning, capturing the spatially continuing information between adjacent GPR traces. In the workflow, some unlabeled GPR images are used to pre-train the SA-DenseCL network for feature extraction and obtaining the backbone weights, which is superior to the conventional pre-training methods of supervised learning pre-trained by ImageNet images. The weights of the pre-trained backbone are then used to initialize the Mask R-CNN through transfer learning. Subsequently, a limited number of labeled GPR images are used to fine-tune the Mask R-CNN for automatically identifying the locations of the reinforcement bars and voids and estimating the secondary lining thickness. The experimental results show that the average precision reaches 96.70%, 81.04%, and 94.67% in identifying reinforcement bar locations, detecting void defects, and estimating secondary lining thickness, respectively, which outperform the conventional methods that use ImageNet-based supervised learning or GPR image-based DenseCL for initializing the Mask R-CNN backbone weights. It is observed that the improved self-supervised learning-based framework can improve the detection and estimation accuracy in GPR tunnel lining inspection.@en