Evaluation of gantry angle optimization for IMRT treatment planning systems using the Pareto front approach

Abstract: Introduction; The aim of this study was to examine the possibilities for MRI-based radiotherapy treatment planning. The work included evaluation of geometric image distortions in MR-images, investigations of how segmentation of bone could be performed from images acquired with the ultra short time of echo (UTE) pulse sequence and assessment of dosimetric differences between treatment planning based on these images as compared to the original CT-based treatment plans.Methods: A phantom with a well-defined rod geometry was used for the studying of geometric distortions in the imaging plane in MR-images. The ability of the UTE pulse sequence to visualize bone and other tissues with short T2-relaxation, was investigated by imaging the head of four healthy volunteers and bone, air and soft tissues were segmented from these images. The dosimetric accuracy expected from such segmented images, as well as from images where the head was treated as one homogenous volume, was compared to the gold standard of using CT-images. The dosimetric evaluation involved nine patients with existing conformal treatment plans and three patients with existing intensity modulated treatment plans.Results: The geometric distortions in the imaging plane were in most cases less than the 2 mm demand in radiotherapy within the investigated volume and with the vendor’s distortion correction applied. Segmentation was feasible using the UTE pulse sequence although some areas were misclassified. For conformal radiotherapy, dose calculations based on the segmented images showed a very close resemblance to the original CT-based plans in regard to dose volume histogram (DVH), dose to planning target volume (PTV) and dose to organ at risk (OAR) although approximating the head as one homogenous volume gave clinically acceptable deviations as well. This trend was also seen for the intensity modulated radiotherapy (IMRT) plans with the same optimization as in the original treatment plan however, making a new optimization based on the derived plans resulted in large deviations in DVH and dose to OAR, possibly a result from misclassifications in the segmentation.Conclusion: Adding the UTE pulse sequence to the T1W- and the T2W-pulse sequences used when imaging patients obviously result in a prolonged scan time (of about 10 min) but has the advantage of making a more accurate treatment planning possible and could lead to MRI being used even in more complicated cases however, more patients are needed for evaluation. To completely fulfill the aim of this study, scanning a patient with both modalities is necessary since it could reveal the performance of the segmentation method and enable a dosimetric evaluation of the MRI-images to be conducted.