Detection of brachytherapy seeds using ultrasound radio frequency signals

Abstract

This paper proposed a novel ultrasonic imaging approach for detecting brachytherapy seeds. Accurate and fast seed localization plays a key role in computing dosimetry for prostate brachytherapy. However, currently used B-mode transrectal ultrasound (TRUS) does not adequately visualize implanted seeds, because the diameter of the seed is quite small and visualization is hampered by speckle noise and angulation of the specular reflection of the seeds. Based on the fact that much more ultrasound wave energy is reflected from metal seeds than from other scatterers in tissue, we developed a new seed detection method directly using ultrasound radio frequency (RF) signals (the raw high frequency echoes before the formation of B-mode TRUS images). It monitors the average power (a version of 2-norm) of the RF signals to measure the reflected wave energy. Each RF scan line is subdivided into a sequence of short segments with the same length and spacing. The average power of each segment is computed by the Fourier based spectra or parametric spectral analysis approaches. In the new method, the logarithmic compression is not applied to the raw RF data, and the average power is proportional to the sum of the square of the signal amplitude. Therefore, it produces significantly higher contrast than conventional B-mode TRUS. Furthermore, the average power algorithm can be implemented very efficiently since no numerical optimization is required. Phantom and ex-vivo experiments show that the average power technique successfully detects implanted brachytherapy seeds, and produces superior results compared with B-mode TRUS imaging.