Sound speed imaging is one modal of ultrasound computed tomography (USCT) which is helpful for early breast disease diagnosis. One of the most critical processes of sound speed reconstruction is time-of-flight picking. As each of the traditional time-of-flight picking methods has shortcomings
for real data, in this study, a practical priorinformation-based combination (PIBC) solution for picking the difference of time-of-flight between the reference data and the object data (DTOF) is proposed to enhance the reconstruction accuracy and uniformity. By using DTOF, some system bias
will be effectively alleviated. Firstly, by analyzing the signal-amplitude, the "penetrating-through-the-object" and the "bypassing-the-object" signals are distinguished. Then for the "penetrating-throughthe-object" signals, based on the 'majority rule,' the consistency of DTOF picked by different
methods are calculated as a basis to combine the advantages of different picking methods; for the "bypassing-the-object" signals, the DTOF closest to zero is chosen. Finally, the DTOFs are post-processed to suppress the noise by a median filter and to fix the deficiency of the system by an
interpolation operator. The new solution is verified by in vitro breast phantom experiment conducted on the home-made USCT system "Lucid." The proposed PIBC solution can quantitatively decrease Root Mean Squared Error (RMSE) and Mean Squared Error (MSE) of DTOF picking and enhance the
image quality of reconstructed sound speed images with higher accuracy and uniformity. This work is significant for ray-based sound speed reconstruction and can provide a fine initial solution for high-resolution wave-based reconstruction.
A Combined Regularization Method Using Prior Structural Information for Sound-Speed Image Reconstruction of Ultrasound Computed Tomography
