Session Item

MR guided radiotherapy is currently finding widespread adoption in clinical practice. The onboard MR images are used for online adaptive radiotherapy treatments. The geometric fidelity of MR images, however, is a potential source of error and should therefore be thoroughly assessed. Since the geometric accuracy is both system and patient dependent, it is important to report the total geometric distortion from the combined effect of gradient inaccuracies and B₀ field inhomogeneities caused by system imperfections and patient-induced susceptibility variations [1]. This abstract reports on the total geometric distortions in 27 patients and seven different treatment sites (prostate, rectum, esophagus, lung, oligo lymph nodes, and breast).

The clinical imaging protocols of 27 patients previously treated or scanned on a 1.5T Elekta Unity system (Elekta AB, Sweden) were used to retrospectively determine the geometric accuracy of these images; Prostate (N=4), Rectum (N=5), Abdominal (N=5), Esophagus (N=1), Lung (N=4), Pelvic lymph nodes (N=5) and Breast (N=2). The protocols consisted of at least a B₀ field map and either a 3D T1-TFE or T2-TSE anatomical image set that was intended for treatment planning purposes. An online visualization tool has been developed to quantify the total geometric distortion from the combined effect of B₀-field inhomogeneities and gradient inaccuracies [2]. This so-called displacer tool was used to assess the patient specific geometric fidelity in the PTV determined as the maximum B₀ induced distortion, the maximum gradient induced distortion as well as the maximum total displacement, defined as the sum of the B₀ and gradient induced distortions. Additionally, the maximum distance between isocenter and PTV was measured.

The results per patient are presented in figure 1a and 1b. The mean results of all patients per treatment site are presented in table 1. For most treatment sites the B₀ induced distortion is less than 0.2 mm with a maximum of 0.5 mm for lung and breast patients. The gradient induced distortion is directly proportional to the maximum isocenter to PTV distance and < 0.7 mm for all treatment sites, excluding two breast patients having a gradient induced distortion of 1 mm. The total displacement within the PTV was well below 1 mm apart from one lung and two breast cancer patients, mainly based on the gradient inaccuracies due to an isocenter distance > 15 cm.

B₀ distortion in most patients is < 0.2 mm and negligible. Gradient inaccuracies were found dominant and directly proportional to isocenter distance (maximum error up to 1 mm). The displacer tool can be used in an online workflow at the MR-Linac for assessment of geometric fidelity and the results can be implemented in the CTV-PTV margin.