ESTRO 2024 - Abstract Book
S3922
Physics - Image acquisition and processing
ESTRO 2024
Figure 1 . The differences in registration of MR to CT comparing observer to clinical values for all patients. Rotations are displayed in A-C, translations in D-F and the magnitude of the difference vectors (based on translations only) are displayed in G. The red line at 1 mm represents the cut-off used to identify patients for extended analysis, while the grey dotted line represents the median difference vector based on all patients. Ten patients had difference vectors equal to or larger than 1 mm. Two patients were excluded due to missing GTV and PTV structures. The maximum CoM shift amongst the eight remaining patients was 4.4 mm, observed for the patient with the largest difference vector (2.1 mm). The DICE Coefficients ranged between 0.61-0.97 and 0.78-0.99, for the GTV and PTVs respectively. The Hausdorff distance ranged from 1.0-5.1 mm, while the median average Hausdorff distance was 0.1 mm.
Conclusion:
The uncertainties associated with the MR-CT image registration in the clinical setting were generally less than 1 mm and the largest differences were observed in the slice direction. However, large variations in the clinical registration may occur for occasional patients, resulting in CoM shifts up to several mm. Although the registration uncertainty is of minor concern for most patients, especially those with large targets combined with generous margins, such as gliomas, it may be of importance when treating small lesions with a few to zero mm margins. Based on this cohort, assuming a 1 mm GTV-PTV margin, 10 out of 45 patients would benefit from MRI-only RT.
Made with FlippingBook - Online Brochure Maker