MRI-US fusion for prostate brachytherapy

Tumour targeting in high dose rate prostate brachytherapy

Long-term disease control for prostate cancer patients can be achieved using high dose rate (HDR) brachytherapy with conventional techniques treating the whole gland. However, such an approach may limit the efficacy of radiotherapy as escalation of dose will be limited by adjacent organs at risk tolerance. Pathology studies suggest that in many cases a dominant cancer focus may exist within the gland and could be at the epicenter of recurrence post treatment. Strategies to identify and intensify treatment to dominant intraprostatic lesions (GTV) are therefore needed, and MRI demonstrates high performance in addressing this need. MRI has been integrated to identify the GTV in order to boost or target intraprostatic lesions.

MRI-US fusion using deformable registration

A significant number of programs are using ultrasound (US) as their treatment planning modality due to its low cost, easy accessibility and real time capability. The current clinical procedures, based on ultrasound images, can’t identify the position of the GTV. Therefore, a spatial registration is needed between MRI and US images in order to accurately position the GTV. Adequate registration between MRI and US would allow dose escalation or focal brachytherapy using an ultrasound technique. On the other hand, most rigid registration algorithms rely on similar prostate shape; which is often inadequate. In fact, the prostate shape is often different between US and MRI images; a transrectal probe is used to obtain US images while MRI images are usually obtained several days before US images, with or without an endorectal coil. In brachytherapy, there is currently no commercially available MRI to US deformable registration algorithm to correct for this difference in prostate shape.

The goal of the project is to create a streamlined workflow to enable clinically feasible MRI-US fusion for prostate HDR brachytherapy. 3D Slicer extension: https://github.com/SlicerRt/SegmentRegistration