Clinical implementation of MRI simulation for IMRT of prostate cancer

L Chen,R.A Price,J Li,L Wang,L Qin,M Ding,C Ma,A Pollack

DOI: https://doi.org/10.1016/S0360-3016(03)01225-2

2003-01-01

Abstract:Purpose/Objective: MR and CT fusion with CT-based dose calculation has become a routine procedure for IMRT treatment planning. The purpose of this study is to investigate the clinical implementation of MRI-based treatment planning (MRI simulation) for prostate cancer. MRI simulation retains the superior soft-tissue contrast for image segmentation while eliminates the potential error in image fusion due to variations in patient position between CT and MRI and uncertainties in localizing the same structures between the two data sets. Furthermore, MRI simulation will avoid redundant CT imaging sessions, which in turn will avoid unnecessary radiation exposure to the patient, reduce treatment cost, and save patient, staff and machine time. Materials/Methods: A 0.23 T open MR scanner (Philips Medical Systems, Cleveland, OH) was used for this study. Patients were scanned in a supine position in an alpha cradle with knee support and foot holder. One axial reference T2-weighted MR image was obtained with three donut-shaped fiducials on the skin to mark the approximate center of the prostate. A saggittal image was obtained through the center of the prostate as depicted on the axial image. The fiducials were then adjusted to mark the isocenter using a set of trackable lasers. The MRI data consisted of 48 slices (3-mm thickness) covering the whole pelvis. Skin marks were then made for daily treatment setup. The MR images were post processed using a gradient distortion correction (GDC) program provided by the vendor. The prostate and other critical structures were contoured and then transferred together with the MRI data to a commercial inverse planning system. Twenty IMRT plans were made using both unit density MRI and CT with the same beam angles, dose constraints and optimization parameters and the resulting plans were compared in terms of isodose distributions and dose volume histograms (DVH). A hybrid plan is generated for both the CT-based plan and the MR-based plan using the leaf sequences and associated MU designed for the corresponding patient plans. The physical phantom is then irradiated using the same leaf sequences to verify the dosimetry accuracy of the treatment plans. Results: Donut shaped fiducial markers worked well in isocenter localization since the center of the fiducial marker could be clearly determined in one MR slice. The CORVUS treatment planning system (version 5.0) allowed dose calculation on both CT and MRI. Small discrepancies (3–5%) in structure volumes were observed between CT and MR due to image resolution and algorithmic differences in volume determination. Small (∼1.5 cm) differences were also found in external contours between CT and MRI due to residual MRI distortion, which resulted in some differences in intensity maps. However, the final dose distributions were equally acceptable based on our clinical criteria (e.g., V95 = 100% of the prescribed target dose, R40 < 35%, R50 < 17% of the rectal volume, etc.). The absolute dose agreed within 3% between measured dose and dose predicted by the planning system in the physical phantom. The BAT ultrasound system was used for daily target localization and a MRI based DRR was used as a reference to assist setup verification. Conclusions: Our results showed that MRI-based simulation meets the accuracy requirement for patient setup, image segmentation, dose calculation and treatment verification for prostate treatment. Commissioning and quality assurance procedures have been developed for MRI simulation and are being implemented clinically for prostate IMRT at this institution.