Session Item

MRI-only radiotherapy (RT) is being developed to eliminate the geometric uncertainty introduced by image registration between MRI and CT data. The technique is advantageous to the patients as well as economically, since it excludes the CT examination. MRI-only RT implementations for prostate cancer have been reported by several clinics, but not yet for brain tumors. Based on a preceding validation study of synthetic CT (sCT) generation for brain, this study aims to prospectively investigate MRI-only brain RT.

Twenty-one glioma patients, including post-surgical cases, were enrolled. The sCT images were generated from MRI Dixon images using the CE-marked deep learning-based software MRI Planner (v 2.2, Spectronic Medical AB, Helsingborg, Sweden). A zero echo time MRI sequence was used to identify the position of the treatment couch. CT scans were acquired but strictly used for sCT quality assurance (QA) only.

The MRI-only workflow included tasks concerning imaging, post-imaging and treatment planning. These were embedded in the treatment planning system and had to be checked before proceeding to the next step in the workflow. RT treatment plans were optimized on sCT with MRI-delineated structures without any influence from the CT images. Dose differences in target and relevant organs at risk between sCT and CT calculated treatment plans were evaluated before treatment was approved. Synthetic CT and CT image qualities were compared through HU line profiles. Retrospective analysis of patient positioning and gamma evaluation of the dose distribution was performed. Results were evaluated with acceptance criteria based on outcomes from the preceding validation study.

This clinical implementation resulted in 20 patients receiving MRI-only RT. The single excluded patient was due to sCT artifacts caused by a hemostatic substance injected near the target during surgery preceding the radiotherapy. All tasks of the workflow could successfully be performed for the treated patients. All dose differences were within ±1 %, comparing dose distributions of the delivered treatment plan (sCT) and the CT-based plan for all patients (Figure 1). Retrospective analysis yielded gamma pass rates (2 %, 2 mm) above 99 %. Patient positioning using CBCT images was within 1 mm for registrations with sCT compared to CT.

Figure 1. Box plot of dose differences (sCT-CT) for the 20 patients treated in the MRI-only workflow.

We report the first successful clinical study of MRI-only radiotherapy for brain tumors, conducted using both prospective and retrospective analysis. Synthetic CT images generated using the CE-marked deep learning-based software for glioma patients with and without anatomical anomalies due to surgery, were clinically robust based on endpoints for dosimetry and patient positioning.