Developing a custom head phantom to validate the accuracy of magnetic resonance imagingguided Radiotherapy

We have developed MRI-only treatment planning, a method of using geometrically corrected MR images for accurate tumor delineation in combination with synthetic CT images derived directly from MR images for more accurate SRS dose calculation and planning. To enter the clinical routine, this procedure must be accurately evaluated and tested as a part of a routine quality control (QC) procedure. The goal of this project is to develop a custom head phantom and use it 1) to validate the accuracy of using MR images for SRS treatment planning and dose delivery; 2) for routine QC.

The standard of care for treating malignant brain tumors includes SRS; the goal is to maximize radiation delivery to tumor tissue while sparing adjacent healthy brain tissue. In the process of radiation treatment planning, MRI or CT images are used for delineating target and healthy tissue. CT imaging is widely available and routinely used for SRS planning, but MRI offers higher soft tissue contrast with the advantage of no ionizing radiation exposure. Thus, MRI improves the accuracy of target delineation, offers better characterization of relevant tumor properties, and reduces the radiation dose to patient.

However, MR images suffer from geometric distortion and a lack of electron density information necessary for dose calculation, which is usually obtained and corrected from and with CT images. Consequently, image registration between MR and CT is necessary. This procedure requires high precision, accuracy, reproducibility and optimized clinical workflow; otherwise, it will cost additional time, money and radiation exposure for the patient, and may introduce inaccuracies to the treatment plan.

Some quality control phantoms are commercially available, but they have major limitations, such as 1) not mimicking brain tissue and air cavities, the major sources of MR image distortion; 2) lacking bone structure, used to validate the accuracy of synthetic CT images; 3) lacking markers and deformable object to check MRI/CT rigid and non-rigid registration; and finally, 4) not offering dosimetric modules (film or ion chamber) to validate the dosimetric accuracy of plans and perform patient-specific QC. Therefore, we propose to develop a MRI-only SRS-appropriate QC phantom, which we hypothesize will increase treatment accuracy and patient safety, save time, and improve clinical outcomes.

In previous efforts toward MRI-only SRS planning, we developed a machine-specific QC procedure for MRI machines and developed a geometric distortion correction algorithm using commercially available phantoms and data from a few patients. In addition, we evaluated the ability of commercially available software, Syngo-Via Frontier (Siemens Medical), to generate synthetic CT images directly from MRI for more accurate SRS planning. Finally, we explored the application of new functional MRI vascular imaging for better tumor delineation and for future post-radiation tumor evaluation. Our goal is to develop and validate a realistic custom head phantom for quality control procedures for MRI- only SRS. The following specific aims are designed to test our hypothesis and extend our initial, promising results:

Aim 1) Assess and improve the geometric accuracy of routine MRI images by combining acquisition and processing methods to minimize residual distortions and validate the effect of correction on SRS planning. Aim 2) Access the geometric and dosimetric accuracy of Synthetic CT images for accurate SRS dose calculation.

Aim 3) Assess both the rigid and non-rigid (deformable) co-registration accuracy of MRI/CT and MRI/MRI images.

Aim 4) Develop an end-to-end system for QC testing for MRI-only SRS planning workflow (imaging/simulation, planning and delivery).

Patent Pending