Waveform inversion of sound speed and acoustic attenuation for ring-array ultrasound tomography based on optimal...

This work investigates a ring-array-based ultrasound computed tomography method for high resolution reconstruction of minute, heterogeneous structures in breast tissue. To address the limitations of conventional ray tomography, which suffers from insufficient resolution due to its neglect of diffraction effects and the strong dependence of waveform inversion on initial models, an inversion framework based on optimal transport theory has been proposed. The full-waveform inversion framework, which employs an objective function based on the 2-Wasserstein metric, improves the global convergence of the inversion process, and helps mitigate the cycle-skipping problem. To further enhance inversion accuracy, a Sigmoid regularization term is incorporated to enforce a nonlinear mapping transformation on the data, which utilizes negative phase information to improve inversion stability. The proposed Sigmoid-regularized optimal transport full-waveform inversion framework achieves high reconstruction accuracy, with maximum relative errors of 2.1% for target diameter, 3.8% for sound speed, and 5.1% for the attenuation coefficient. Experimental validation demonstrated that the reconstructed results exhibited high anatomical consistency with B-mode ultrasound, and the spatial matching degree of lesion areas reached 92.6% as validated by the Dice coefficient. This approach enables high-precision multimodal imaging for clinical applications.