Search

An algorithm was developed for the correction of ring artifacts in phase-insensitive ultrasound computed tomography attenuation images. Differences in the measurement sensitivity between the ultrasound transducer array elements cause discontinuities in the sinogram which manifest as rings and arcs in the reconstructed image. The magnitudes of the discontinuities are potentially time-varying and dependent on the attenuation being measured. The algorithm dynamically determines the measurement sensitivity of each transducer in the array during the scan by comparison with both the elements to its left and the elements to its right. Elements at either end of the array are corrected, assuming a zero-attenuation path. The two estimates of sensitivity are combined using a weighted mean similar to a Kalman filter. The algorithm was tested on simulated and experimentally acquired data. It was demonstrated to reduce the root-mean-square error (RMSE) of simulated images against ground-truth images by up to a factor of 50 compared with uncorrected images and to visibly reduce artifacts on images reconstructed from the experimentally acquired data. [ABSTRACT FROM AUTHOR]

This paper describes the application of a pulse pileup correction algorithm based on maximum likelihood estimation to simulated and experimentally acquired signals from a pyroelectric ultrasound sensor. The sensor forms part of a prototype phase-insensitive ultrasound computed tomography (piUCT) system for breast cancer detection that generates quantitative maps of acoustic attenuation. The effectiveness of the piUCT technique has been previously demonstrated through imaging of cylindrical, polyurethane phantoms. For the technique to be applied clinically, the system must be capable of measuring high acoustic attenuations (>30 dB) and completing a scan in a clinically acceptable time frame of approximately 5 min. High attenuations present a challenge because the sensor responds directly to acoustic power and therefore, an attenuation of 30 dB causes a drop in signal level of three orders of magnitude, resulting in a large dynamic range in the received signal set. Completion of a scan in 5 min requires a period between measurements shorter than the response time of the sensor, causing pyroelectric pulse pileup. The maximum likelihood estimation method recovers the amplitudes of closely spaced superimposed pyroelectric signals while improving the coefficient of variation of the measurement by a factor of two compared to direct pulse amplitude measurement, simultaneously increasing the maximum measurable attenuation and measurement rate of the system. [ABSTRACT FROM AUTHOR]