Deep learning for ultrasound image formation is rapidly garnering research support and attention, quickly rising as the latest frontier in ultrasound image formation, with much promise to balance both image quality and display speed. Despite this promise, one challenge with identifying optimal solutions is the absence of unified evaluation methods and datasets that are not specific to a single research group. This article introduces the largest known international database of ultrasound channel data and describes the associated evaluation methods that were initially developed for the challenge on ultrasound beamforming with deep learning (CUBDL), which was offered as a component of the 2020 IEEE International Ultrasonics Symposium. We summarize the challenge results and present qualitative and quantitative assessments using both the initially closed CUBDL evaluation test dataset (which was crowd-sourced from multiple groups around the world) and additional in vivo breast ultrasound data contributed after the challenge was completed. As an example quantitative assessment, single plane wave images from the CUBDL Task 1 dataset produced a mean generalized contrast-to-noise ratio (gCNR) of 0.67 and a mean lateral resolution of 0.42 mm when formed with delay-and-sum beamforming, compared with a mean gCNR as high as 0.81 and a mean lateral resolution as low as 0.32 mm when formed with networks submitted by the challenge winners. We also describe contributed CUBDL data that may be used for training of future networks. The compiled database includes a total of 576 image acquisition sequences. We additionally introduce a neural-network-based global sound speed estimator implementation that was necessary to fairly evaluate the results obtained with this international database. The integration of CUBDL evaluation methods, evaluation code, network weights from the challenge winners, and all datasets described herein are publicly available (visit https://cubdl.jhu.edu for details).

Design, embodiment, and experimental study of a novel concept of extracorporeal phased array ultrasound transducer for prostate cancer regional deep hyperthermia treatments using a transperineal acoustic window is presented. An optimized design of hyperthermia applicator was derived from a modelling software where acoustic and thermal fields were computed based on anatomical data. Performance tests have been experimentally conducted on gel phantoms and tissues, under 3T MRI guidance using PRFS thermometry. Feedback controlled hyperthermia (ΔT = 5 °C during 20min) was performed on two ex vivo lamb carcasses with prostate mimicking pelvic tissue, to demonstrate capability of spatio-temporal temperature control and to assess potential risks and side effects. Our optimization approach yielded a therapeutic ultrasound transducer consisting of 192 elements of variable shape and surface, pseudo randomly distributed on 6 columns, using a frequency of 700 kHz. Radius of curvature was 140 mm and active water circulation was included for cooling. The measured focusing capabilities covered a volume of 24 × 50 × 60 mm³. Acoustic coupling of excellent quality was achieved. No interference was detected between sonication and MR acquisitions. On ex vivo experiments the target temperature elevation of 5 °C was reached after 5 min and maintained during another 15 min with the predictive temperature controller showing 0.2 °C accuracy. No significant temperature rise was observed on skin and bonny structures. Reported results represent a promising step toward the implementation of transperineal ultrasound hyperthermia in a pilot study of reirradiation in prostate cancer patients.

Ultrasound is typically measured using phase-sensitive piezoelectric sensors. Interest in phase-insensitive sensors has grown recently, with proposed applications including ultrasound attenuation tomography of the breast and acoustic power measurement. One advantage of phase-insensitive detectors, in contrast to conventional phase-sensitive detectors, is that they do not suffer from a narrow directional response at high frequencies due to phase cancellation. A numerical model of a phase-insensitive pyroelectric ultrasound sensor is presented. The model consists of three coupled components run in sequence: acoustic, thermal, and electrical. The acoustic simulation models the propagation and absorption of the incident ultrasound wave. The absorbed acoustic power density is used as a heat source in the thermal simulation of the time-evolution of the temperature in the sensor. Both the acoustic and thermal simulations are performed using the k-Wave MATLAB toolbox with an assumption that shear waves are not supported in the medium. The final component of the model is a pyroelectric circuit model which outputs the sensor response based on the temperature change in the sensor. The modelled pyroelectric sensor response and directional dependence are compared to empirical data.

The aim of this study was to assess the accuracy of a custom-developed numerical simulation of hepatic multibipolar radiofrequency ablation (mbRFA) under standardized conditions, accounting for temperature-dependent changes in tissue properties, dehydration and coagulation. The model was specifically assessed for its ability to predict vascular cooling effects induced by large vessels using an established ex vivo porcine liver model. A GPU-accelerated finite-difference time-domain (FDTD) simulation coupling the quasi-static Maxwell and Pennes' bioheat equations was implemented to model mbRFA. Ex vivo mbRFA were performed using three internally cooled bipolar applicators. A saline-perfused glass tube was employed to simulate a standardized blood flow and positioned at five predefined distances from the ablation center in 2.5 mm increments. Simulated and experimental ablations were compared using area-, surface- and classification-based metrics on a standardized two-dimensional (2D) cross section to evaluate the model's predictive accuracy. The numerical simulation reliably predicted the ablation shape for mbRFA while effectively accounting for vascular cooling effects. Conformity between simulated and real ablation areas was 96% across all experimental settings. The average similarity between the ex vivo ablations and the simulation were 0.92 (Dice coefficient) and 0.85 (Jaccard coefficient), respectively. Compared to the reference area, the simulation overestimated the ablation zone by 4% (false-positive regions) and underestimated it by 10% (false-negative regions). Overall, the numerical model demonstrated good spatial agreement with the experimental ablations, confirming its suitability for quantitative prediction of ablation geometry.

Radiofrequency (RF) ablation refers to a minimally invasive tumour ablation treatment using RF electromagnetic waves. A needle is placed inside the tumour, and an electrical current is applied, which is heating the tissue to burn the tumour. For treatment planning, the heat propagation is simulated, but the treatment volumes are significantly affected by the tissue properties, which vary between different patients, based upon both the vasculature and levels of fat and water content. Undertreatment can lead to tumour recurrence, while over‐treatment damages healthy tissue. We propose to study the effect of tissue properties on the ablation based on an interactive visual analysis of simulation ensembles, where the tissue properties form the parameter space of the ensemble. The proposed Tissue Property Analysis Tool (TPAT) uses 2D and 3D spatial visualizations for comparative analysis of simulation outcomes for different parameter settings. A 3D parameter‐space visualization allows for the analysis of the effect on the ablation result when modifying a selected parameter for the three involved tissues (tumour, liver and vessels). During the analysis, any parameter setting shall be accessible. When no simulation outcome has been generated for a selected parameter setting, we deploy a deep‐learning‐based surrogate model to predict an ablation outcome. Using our deep learning approach, we achieve an improved prediction accuracy both within and across spatial configurations, which outperforms interpolation‐based schemes in 94.32% and 99.07% of the test data, respectively. We discuss our approach with domain experts for developing simulation models and demonstrate the usefulness of our approach for analysing the effect of tissue properties on RF ablation of liver tumours.