Localization of linked 125I seeds in postimplant TRUS images for prostate brachytherapy dosimetry

Purpose: To demonstrate that 125I seeds can be localized in transrectal ultrasound (TRUS) images obtained with a high-resolution probe when the implant is performed with linked seeds and spacers. Adequate seed localization is essential to the implementation of TRUS-based intraoperative dosimetry for prostate brachytherapy. Methods and Materials: Thirteen preplanned peripherally loaded prostate implants were performed using 125I seeds and spacers linked together in linear arrays that prevent seed migration and maintain precise seed spacing. A set of two-dimensional transverse images spaced at 0.50-cm intervals were obtained with a high-resolution TRUS probe at the conclusion of the procedure with the patient still under anesthesia. The image set extended from 1.0 cm superior to the base to 1.0 cm inferior to the apex. The visible echoes along each needle track were first localized and then compared with the known construction of the implanted array. The first step was to define the distal and proximal ends of each array. The visible echoes were then identified as seeds or spacers from the known sequence of the array. The locations of the seeds that did not produce a visible echo were interpolated from their known position in the array. A CT scan was obtained after implantation for comparison with the TRUS images. Results: On average, 93% (range, 86–99%) of the seeds were visible in the TRUS images. However, it was possible to localize 100% of the seeds in each case, because the locations of the missing seeds could be determined from the known construction of the arrays. Two factors complicated the interpretation of the TRUS images. One was that the spacers also produced echoes. Although weak and diffuse, these echoes could be mistaken for seeds. The other was that the number of echoes along a needle track sometimes exceeded the number of seeds and spacers implanted. This was attributed to the overall length of the array, which was approximately 0.5 cm longer than the center-to-center distance between the first and last seed owing to the finite length of the seeds at the ends of the array. When this occurred, it was necessary to disregard either the most distal or most proximal echo, which produced a 0.5-cm uncertainty in the location of the array in the axial direction. For these reasons, simply localizing the visible echoes in the TRUS images did not guarantee the reliable identification of the seeds. Conclusion: Our results have demonstrated that a high p