A quantitative ultrasound imaging system for generating high-resolution 2D and 3D maps of physical properties

Background

Ultrasound imaging is a widely used diagnostic tool in medicine and non-destructive testing, valued for its real-time imaging capabilities, safety, and relatively low cost. Traditional ultrasound systems employ arrays of transducers to emit and receive acoustic waves, constructing images based on the echoes reflected from internal structures. The most common technique, B-mode imaging with delay-and-sum processing, provides qualitative images that help visualize anatomical features and detect abnormalities. However, these images are primarily qualitative: the brightness or intensity of a pixel does not directly correspond to a physical property of the tissue or material. As a result, clinicians and engineers often face challenges in making precise, quantitative assessments of tissue health or material integrity, highlighting a growing need for imaging methods that can provide spatially resolved, physically meaningful maps of properties like wavespeed, density, and elasticity.

Current approaches to quantitative ultrasound imaging, such as elastography and diffraction tomography, have notable limitations. Elastography techniques, which estimate tissue elasticity by tracking the propagation of shear waves or tissue displacement, are often sensitive to noise and provide limited physical interpretability, especially in heterogeneous or complex media. Diffraction tomography, while capable of reconstructing quantitative maps under certain conditions, requires restrictive data acquisition geometries and is effective only when the contrast in wavespeed is low, limiting its practical applicability. Time-reversal and related nonlinear imaging methods can improve the detection of small scatterers but are generally restricted to point-like targets and require prior knowledge of interface locations. Overall, these limitations prevent existing methods from reliably producing high-resolution, quantitative 2D or 3D images of physical parameters in a broad range of clinical and industrial scenarios, underscoring the need for more robust and accurate quantitative imaging solutions.

Technology

This technology is a quantitative ultrasound imaging solution that applies seismic Full Waveform Inversion (FWI) principles to generate high-resolution 2D or 3D maps of physical properties such as wavespeed, density, and elasticity within a medium. The process begins by emitting ultrasound signals into the target area, with arrays of transducers capturing both transmitted and reflected waves. These observed signals are then compared to synthetic data generated by numerically simulating wave propagation through an initial model of the medium. Using an iterative optimization framework, the system updates the model parameters to minimize the difference between observed and simulated data, employing advanced cost functions—such as waveform, instantaneous phase, and double difference—to enhance robustness and accuracy. The method leverages GPU-accelerated simulations, multi-frequency scaling, and regularization techniques like Gaussian smoothing, enabling efficient and precise reconstruction of the medium’s internal structure, even in complex or noisy environments.

What differentiates this technology is its ability to produce quantitative, physically meaningful images rather than traditional qualitative ultrasound images. Unlike conventional B-mode ultrasound, which is limited by artifacts, lack of direct physical interpretation, and qualitative brightness mapping, this solution reconstructs spatially resolved maps of intrinsic material properties. The use of seismic FWI—adapted for ultrasound—enables the system to handle both continuous and discontinuous structures, accommodate various acquisition geometries, and remain robust to noise and practical constraints. The integration of sophisticated cost functions and GPU acceleration ensures rapid convergence and high-resolution results, making the technology suitable for demanding applications in medical diagnostics and non-destructive testing where accurate characterization of internal structures is critical.

Benefits

• Generates high-resolution, quantitative 2D and 3D maps of physical properties such as wavespeed, density, and elasticity, enabling physically interpretable imaging.
• Improves accuracy and robustness over traditional ultrasound imaging by minimizing artifacts like speckle and noise through full waveform inversion.
• Supports multiple cost functions (waveform, instantaneous phase, double difference, hybrid) to enhance convergence and reduce sensitivity to noise and modeling errors.
• Utilizes GPU-accelerated simulations and multi-scaling frequency strategies to achieve practical computation times for complex 2D and 3D reconstructions.
• Flexible data acquisition setups, including one-sided and two-sided transducer arrays and submersion in liquid, allow precise control and modeling of the imaging environment.
• Applicable to medical diagnostics and non-destructive testing by providing quantitative, physically meaningful characterization of tissues or materials.
• Employs advanced regularization and optimization techniques to avoid local minima and ensure stable inversion results without manual intervention.

Applications

• Medical tissue characterization
• Non-destructive material testing
• 3D quantitative elastography
• Industrial defect detection
• Advanced biomedical research imaging

Additional info

This quantitative ultrasound imaging method utilizes seismic Full Waveform Inversion to generate 2D or 3D maps of wavespeed, density, and elasticity. It iteratively compares observed acoustic data against synthetic simulations using numerical wave solvers. By updating a physical model via the adjoint state method and GPU acceleration, the system minimizes residuals through specific cost functions to reconstruct high-resolution, quantitative images of internal structures.