Testing and evaluation of image registration brachytherapy

Abstract

Real-time ultrasound (US) is used to guide implantation for temporary and permanent prostate brachytherapy treatment. MRI, CT and US fusion is available in dedicated treatment planning systems used for these procedures. The registration of these imaging modalities requires additional image verification and tests to ensure accuracy. Using a CIRS Tissue Equivalent Ultrasound Prostate Phantom we evaluated the accuracy of the registration system of a Hitachi Preirus US scanner for fusing MRI and live US images. The phantom contains 3 lesions, a urethra and a prostate volume of 53 cm3. The MRI and US images were registered using the urethra as a marker in the central slice of the image. The registration was then checked at the base and apex of the prostate. Along the urethra the registration was accurate to within 2mm. Another common fusion software, Variseed™, was assessed by measuring the percentage of overlapping volumes for the prostate and a lesion within the prostate for CT, US and MRI images. The overlapping volumes were compared for CT–MRI, CT–US (un-deformed prostate), CT–US (deformed prostate), MRI-US (un-deformed prostate), MRI-US (deformed prostate) and US–US(un-deformed prostate and deformed prostate). Contour overlap varied from 90% to 40% for the lesion, which was approximately 1 cm3, and 96% to 78% contour overlap for the whole prostate. This study demonstrates how a tissue equivalent phantom can be used for testing and commissioning of multi-modality registration systems in brachytherapy.