Session Item

Vendor-approved Magnetic Resonance (MR) sequences are provided for MR-linac workflows however alternative ‘off-label’ sequences may offer advantages. Prior to clinical use, the safety, accuracy and overall clinical benefit should be determined. The aim of this study was to develop a systematic approach for off-label sequence evaluation and demonstrate the application for bladder cancer MR-guided radiotherapy (MRgRT) on Unity (Elekta AB, Stockholm).

Two T2-weighted Turbo Spin Echo off-label sequences were proposed, denoted 1.1min3mm and 3.4min2mm, indicating both acquisition time and slice thickness. Each sequence was assessed against the vendor provided 2min1mm sequence.

The off-label sequences were acquired during opportunity scanning time on 3 patients receiving daily bladder cancer MRgRT, for up to 3 fractions per patient. Images were exported to Monaco treatment planning system (TPS) for offline review (v5.40.1, Elekta). Microsoft forms employing four-point Likert scales were created to guide qualitative review of image quality and workflow suitability, carried out independently by two doctors and two treatment radiographers.

Geometric accuracy, image uniformity and ghosting were assessed using the ACR phantom (acr.org). The impact of the larger slice thickness within the TPS included assessment of margin expansions, optimisation behaviour and dosimetric impact (carried out by physicists). Existing clinical bladder treatment image data (2min1mm) were used in VolumeView (v5.3.31, Philips Medical Systems, Best) to reconstruct 2mm and 3mm slice thickness data sets for plan generation using the same clinical template. Treatment plans were then recalculated on the corresponding 2min1mm sequence and were compared using clinical goals, and dose conformity.

To quantify the potential reduction in overall treatment time for the shorter acquisition of the 1.1min3mm sequence, a phantom treatment workflow was carried out.

Table 1 summarises the results. Both sequences were shown to be accurate, safe and of non-inferior utility to the 2min1mm sequence and facilitated clinically acceptable treatment plans in Monaco. For margins that are not integer multiples of the slice thickness, the effective margin can differ (see Fig 1), however no dosimetric impact was seen. The workflow time reduction of 4min for the 1.1min3mm sequence can be attributed to both shorter acquisition and reduced dataset size. This could translate into a 14% time reduction to the average bladder cancer MRgRT at our centre.

A systematic sequence evaluation was developed and used to evaluate off-label sequences for use in bladder MRgRT on the Unity MR-Linac. The use of the proposed 1.1mm3mm sequence will enable meaningful reduction of the workflow time. This has potential to minimise intra-fraction target volume changes and permit safe future PTV margin reduction. This approach to evaluating off-label sequences will be applied to other indications and sequences before clinical implementation.