Search

In recent years, synthetic Computed Tomography (CT) images generated from Magnetic Resonance (MR) or Cone Beam Computed Tomography (CBCT) acquisitions have been shown to be comparable to real CT images in terms of dose computation for radiotherapy simulation. However, until now, there has been no independent strategy to assess the quality of each synthetic image in the absence of ground truth. In this work, we propose a Deep Learning (DL)-based framework to predict the accuracy of synthetic CT in terms of Mean Absolute Error (MAE) without the need for a ground truth (GT). The proposed algorithm generates a volumetric map as an output, informing clinicians of the predicted MAE slice-by-slice. A cascading multi-model architecture was used to deal with the complexity of the MAE prediction task. The workflow was trained and tested on two cohorts of head and neck cancer patients with different imaging modalities: 27 MR scans and 33 CBCT. The algorithm evaluation revealed an accurate HU prediction (a median absolute prediction deviation equal to 4 HU for CBCT-based synthetic CTs and 6 HU for MR-based synthetic CTs), with discrepancies that do not affect the clinical decisions made on the basis of the proposed estimation. The workflow exhibited no systematic error in MAE prediction. This work represents a proof of concept about the feasibility of synthetic CT evaluation in daily clinical practice, and it paves the way for future patient-specific quality assessment strategies.

Abstract Background and purpose Rare but severe toxicities of the optic apparatus have been observed after treatment of intracranial tumours with proton therapy. Some adverse events have occurred at unusually low dose levels and are thus difficult to understand considering dose metrics only. When transitioning from double scattering to pencil beam scanning, little consideration was given to increased dose rates observed with the latter delivery paradigm. We explored if dose rate related metrics could provide additional predicting factors for the development of late visual toxicities. Materials and methods Radiation-induced intracranial visual pathway lesions were delineated on MRI for all index cases. Voxel-wise maximum dose rate (MDR) was calculated for 2 patients with observed optic nerve toxicities (CTCAE grade 3 and 4), and 6 similar control cases. Additionally, linear energy transfer (LET) related dose enhancing metrics were investigated. Results For the index cases, which developed toxicities at low dose levels (mean, 50 GyRBE), some dose was delivered at higher instantaneous dose rates. While optic structures of non-toxicity cases were exposed to dose rates of up to 1 to 3.2 GyRBE/s, the pre-chiasmatic optic nerves of the 2 toxicity cases were exposed to dose rates above 3.7 GyRBE/s. LET-related metrics were not substantially different between the index and non-toxicity cases. Conclusions Our observations reveal large variations in instantaneous dose rates experienced by different volumes within our patient cohort, even when considering the same indications and beam arrangement. High dose rate regions are spatially overlapping with the radiation induced toxicity areas in the follow up images. At this point, it is not feasible to establish causality between exposure to high dose rates and the development of late optic apparatus toxicities due to the low incidence of injury.