Speed-of-sound validation in ring array ultrasound computed tomography using transmission-mode tissue-mimicking phantoms.
Abstract
Ultrasound computed tomography (USCT) is a promising radiation-free quantitative imaging technique for soft tissue characterization, yet the absence of standardized validation phantoms hinders its clinical standardization and algorithm benchmarking. In this study, a suite of agar-based tissue-mimicking phantoms was developed and fabricated for the speed-of-sound (SoS) validation of a ring-array USCT system. The phantoms featured a stable background medium (SoS of 1540 m/s) and embedded cylindrical inclusions of 5 mm and 10 mm diameters with three clinically relevant target SoS values (1475, 1500, and 1510 m/s against a 1540 m/s background). The SoS of all materials was independently characterized by through-transmission measurements, with their SoS confirmed stable by repeated testing over two years. A clinical 2,048-element ring-array USCT system was adopted for data acquisition, and three reconstruction methods (ray-based method, L2-norm full-waveform inversion (L2-FWI), and UNN-FWI) were quantitatively compared. Experimental results demonstrated that UNN-FWI with implicit regularization outperformed conventional methods, exhibiting the lowest area and SoS reconstruction errors for all SoS differences and target sizes. L2-FWI suffered from cycle-skipping artifacts and obvious quantitative underestimation, while ray-based methods showed severe spatial resolution limitations. This work provides a standardized, reusable phantom-based validation framework for USCT SoS imaging. The publicly released datasets further support global algorithm benchmarking, filling the critical gap in USCT quantitative quality assurance and facilitating the clinical translation and standardization of USCT technology.