In vivo 3D ultrasound computed tomography of musculoskeletal tissues with generative neural physics

2025-08-16

AI Summary

Traditional ultrasound computed tomography (USCT) for musculoskeletal imaging relies on ray-based approximations, failing to model strong scattering effects—leading to low quantitative accuracy in acoustic parameter estimation and slow reconstruction. This work proposes a generative neural-physics framework that synergistically integrates full-waveform inversion with physics-informed neural networks to construct a compact, data-efficient surrogate model. Leveraging cross-modal image priors, it enables high-fidelity 3D reconstruction of acoustic parameters—sound speed and attenuation—in strongly scattering media using only sparse measured data. The method achieves MRI-level spatial resolution (~1 mm) for in vivo 3D quantitative imaging of muscle and bone tissues within 10 minutes, enabling, for the first time, radiation-free, biomechanically sensitive ultrasonic tomographic visualization. Validation across in vivo datasets—including breast, upper-limb, and lower-limb acquisitions—demonstrates substantial improvements over conventional USCT approaches.

Technology Category

• Computer Vision: Low Level & Physics-based Vision
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Web Mining and Content Analysis: Web data generation and simulation

Abstract

Ultrasound computed tomography (USCT) is a radiation-free, high-resolution modality but remains limited for musculoskeletal imaging due to conventional ray-based reconstructions that neglect strong scattering. We propose a generative neural physics framework that couples generative networks with physics-informed neural simulation for fast, high-fidelity 3D USCT. By learning a compact surrogate of ultrasonic wave propagation from only dozens of cross-modality images, our method merges the accuracy of wave modeling with the efficiency and stability of deep learning. This enables accurate quantitative imaging of in vivo musculoskeletal tissues, producing spatial maps of acoustic properties beyond reflection-mode images. On synthetic and in vivo data (breast, arm, leg), we reconstruct 3D maps of tissue parameters in under ten minutes, with sensitivity to biomechanical properties in muscle and bone and resolution comparable to MRI. By overcoming computational bottlenecks in strongly scattering regimes, this approach advances USCT toward routine clinical assessment of musculoskeletal disease.

Problem

• Overcoming limitations of conventional USCT for musculoskeletal imaging
• Enabling fast, high-fidelity 3D ultrasound computed tomography
• Improving quantitative imaging of in vivo musculoskeletal tissues

Innovation

• Generative neural physics framework for 3D USCT
• Learns ultrasonic wave propagation from cross-modality images
• Reconstructs 3D tissue parameter maps in under ten minutes