Program Information
Optimization of Synthetic CT for Female Pelvis
J Kim*, A Ghanem , M Elshaikh , I Chetty , C Glide-Hurst , Henry Ford Health System, Detroit, MI
Presentations
MO-RAM-GePD-JT-2 (Monday, July 31, 2017) 9:30 AM - 10:00 AM Room: Joint Imaging-Therapy ePoster Theater
Purpose: While MR-only treatment planning has become available for male pelvis, female synthetic CT (synCT) solutions have not been widely implemented. This work evaluates the translation of a previously validated male synCT solution to female pelvis taking site-specific anatomical and planning considerations into account.
Methods: MR simulation data for 7 female pelvis patients were acquired on a 1.0T MR-SIM scanner. Four patients were treated using 3D planning, and three were treated using VMAT. A single physician contoured OARs (bladder and rectum), uterus, and bony anatomy on T2-weighted images according to GEC-ESTRO guidelines. T2 and T1-weighted images were used for synCT generation. First, synCTs were generated using previously optimized male pelvis weights. Then, our synCT solution was reoptimized to determine the need for female-specific weights and quantify differences. Optimization was performed by minimizing the L2 norm between synCT and simulation CT values. Mean absolute error (MAE) was calculated for each set of synCTs. Clinical treatment plans were copied to both synCTs for each patient and recalculated using original MUs.
Results: Between optimizations, weights were consistent (<5% difference) for fat and fluid. For bone and soft tissue, female T1 weights increased significantly (>15%) with a corresponding decrease for T2 and inverse T1 weights. Nevertheless, mean MAE across all patients was 88.6+16.1 HU for male-optimized weights, and 90.7+17.1 HU for weights optimized using the female patients. Dosimetric differences between plans calculated on different synCTs were small (<0.5%). 3D patients were slightly more sensitive than VMAT patients due to differences in bone weighting.
Conclusion: SynCT weighting differences produced small changes for MAE and calculated dose distributions, suggesting that male pelvis weights are good approximations of female data. However, 3D plans with beams traversing through bony anatomy may require more robust bone solutions. Future work will involve incorporating female pelvis data into a brachytherapy workflow.
Funding Support, Disclosures, and Conflict of Interest: Research partially supported by the NCI of NIH under Award #R01CA204189.
Contact Email: