MO-FG-CAMPUS-JeP2-04: Comparison Study for CT and MR-Only Prostate IMRT Treatment Planning: A Framework for the Estimation of Relative Contribution of Body Contour Discrepancies Tissue Stratification and HU-RED Conversion to the Overall Dose Difference

Maspero, M, Seevinck, P R, Schubert, G, Hoesl, M, Meijer, Gert, Viergever, M A, Lagendijk, Jjw, van den Berg, Cat


Medical Physics 43 (6), p. 3721-3722


PURPOSE: To compare dose of prostate IMRT plans on CT vs MR-based pseudo-CT (pCT). A framework is demonstrated to isolate various confounding factors causing dose differences. This approach will be valuable for medical physicists interested in commissioning strategy of MR-only dose planning.

METHODS: A retrospective study was conducted on fourteen prostate patients that underwent CT and MRI scans. MR was equipped with an in-house-developed flat table-top and coil support. Time between both scans was 1-2 hours.MR-based pCTs (MRpCTs) were generated from a dual-echo GRE Dixon sequence (<2.5 min) using a hybrid automatic algorithm combining model based segmentation and HU bulk-assignment (MRCAT prototype Philips). For each patient CTs were registered to MRpCT and three different CT-based pCTs were obtained, allowing for separation of dose differences due to bulk-density tissue stratification and body contour discrepancies. Clinical 77 Gy 10 MV IMRT plans were simulated on the registered-CT and re-calculated on each pCT using two commonly-used HU-RED calibration curves. Dose differences maps were compared for CTV and VBody.

RESULTS: Averaged (±1std) over the patient population MRpCTs in the CTV resulted in 0.33±0.39% higher dose compared to the registered-CT employing the same HU-RED calibration curves. On the same datasets the difference was 0.66±0.21% when using different calibration curves. Stratification applied to CT-based pCT introduced differences of 0.17±0.13% whereas dose differences due to mismatch of body contour was 0.01±0.29%.

CONCLUSION: This study evaluated the accuracy of dose calculation on MR-only generated pCT for RTP resulting in dose differences which were not statistically significant (within 1 std). The relative contribution of possible confounding factors to the total dose difference was quantified: HU-RED conversion was demonstrated to contribute relatively the most followed by bulk density tissue stratification while body contour differences appeared to average out over the patient population. The research is founded by ZonMw IMDI Programme project number: 10-104003010. The project is co-founded by Philips Healtcare. G. Schubert is employed by Philips Healthcare Vantaa Finland.