End-to-end phantom evaluation for transition from MR/CT combined to MR-Only workflow.
Cunningham JM, and Glide-Hurst C. End-to-end phantom evaluation for transition from MR/CT combined to MR-Only workflow. Med Phys 2017; 44(6):2818.
Purpose: As we move toward MR-only treatment planning, a need exists to distinguish discrepancies between MR-only and MR/CT combined workflows. We evaluated a new MR-safe stereotactic head phantom for its performance for end-to-end testing for MR-SIM. Methods: We require a phantom that can mimic the entire MR-SIM workflow including: setup in site-specific coils, data acquisition using the same sequences, isocenter marking, synthetic CT generation or co-registration for MR/CT combined, planning, and delivery with dosimetric measurements. The head phantom was immobilized using an MR-compatible headrest and external MR/CT fiducials were aligned to the magnet laser system. 1.0T brain MR-SIM scans (T1-weighted, T2- weighted, FLAIR, and UTE-Dixon) were acquired using an 8-channel head coil. An MRI/CT target insert (12.5 cc volume) was filled with signal generator fluid and embedded in a ∼380 cc water-filled cube insert. CT-SIM was conducted using the same setup. Images were transferred to the treatment planning system and rigid registration was performed. Results: The head phantom was challenging to align in MR-SIM due to the head coil and irregular posterior phantom surface. While the phantom includes a custom-made cradle, it would not fit in the head coil. Setup was more straightforward in CT-SIM. MR/CT compatible fiducial landmarks were visible in both modalities. The target insert was visible and exhibited contrast from the water-filled cube. Co-registered images illustrated strong agreement between MR and CT images. However, much of the phantom did not yield visible MR signal, suggesting that generation of a synthetic CT is not possible. In addition, dosimetric measurements would be challenging to obtain in the fluid-filled inserts. Conclusion: The MR/CT compatible phantom may evaluate the geometric integrity of simulation although it will not test all of the unique requirements for MR-only planning. Further development is warranted as we move toward an entirely MR-based workflow including MR-IGRT.