Quantitative analysis of MRI‐guided radiotherapy treatment process time for tumor real‐time gating efficiency

Abstract Purpose Magnetic Resonance‐guided radiotherapy (MRgRT) systems allow continuous monitoring of therapy volumes during treatment delivery and personalized respiratory gating approaches. Treatment length may therefore be significantly affected by patient’s compliance and breathing control. We quantitatively analyzed treatment process time efficiency (TE) using data obtained from real‐world patient treatment logs to optimize MRgRT delivery settings. Methods Data corresponding to the first 100 patients treated with a low T hybrid MRI‐Linac system, both in free breathing (FB) and in breath hold inspiration (BHI) were collected. TE has been computed as the percentage difference of the actual single fraction’s total treatment time and the predicted treatment process time, as computed by the TPS during plan optimization. Differences between the scheduled and actual treatment room occupancy time were also evaluated. Finally, possible correlations with planning, delivery and clinical parameters with TE were also investigated. Results Nine hundred and nineteen treatment fractions were evaluated. TE difference between BHI and FB patients’ groups was statistically significant and the mean TE were 42.4%, and −0.5% respectively. No correlation was found with TE for BHI and FB groups. Planning, delivering and clinical parameters classified BHI and FB groups, but no correlation with TE was found. Conclusion The use of BHI gating technique can increase the treatment process time significantly. BHI technique could be not always an adequate delivery technique to optimize the treatment process time. Further gating techniques should be considered to improve the use of MRgRT.


| INTRODUCTION
Cancer is one of the leading causes of death throughout the world.
Each year, 4.6 million new cancer cases are diagnosed in the WHO European Region and 2.1 million people die from cancer. 1 Radiotherapy (RT) treatment is always playing always a greater role and RT cancer treatments have been developed from relatively simple processes into very complex procedures, recently introducing several new technologies and delivery techniques into clinical practice. 2 One of the most innovative technologies is represented by Magnetic Resonance-guided Radiotherapy (MRgRT) hybrid units, that combine high dose distribution conformality and online adaptation with high quality positioning imaging. 3,4 The MRIdian Linac system (ViewRay Inc., Mountain View, California, US) has been the first example of a hybrid RT machine authorized for clinical treatments, joining a 0.35 Tesla MRI on board scanner with a 6 MV flattening filter free (FFF) Linac system. 3 One of the most significant advantages of this technology is represented by the possibility to monitor online the target during the whole radiotherapy treatment, using sagittal MR images acquired in cine modality with 4 frames/s. This leads to evident advantages especially in the case of the irradiation of movable targets, successfully integrating the motion management strategies to date available in current radiotherapy such as respiratory gating with surface markers or surface motions control. 5 However, these methods are exposed to intra-fractional changes in the relationship between the internal tumor/OARs motion and the external surface related signal. 6,7 To reduce the uncertainties of the issue of gating with a surrogate signal, internal markers could be implanted inside or close to the target lesion: nevertheless, this process is invasive for the patient and time consuming.
Thanks to the use of a real-time sagittal MRI acquisition with a temporal frequency of 4 frames per second, the target is irradiated, as shown in Fig. 1, only when it is located within a predefined boundary, usually defined as a geometric expansion of the target structure itself. 8 During treatment simulation, the breathing modality is also evaluated, and the most appropriate delivery technique is chosen between breath hold (BH) or free breathing (FB): this decision is generally treatment site dependent. Treatment sites where breathing motion leads to a relevant variation of the lesion position, such as lung, liver, pancreas, adrenal gland and (some) lymph nodes are usually treated in BH. Nevertheless, for example, soma central lung lesions can be treated in free breathing since the lesion excursion due to the breathing path is negligible. On the other hand, treatment sites where breathing motion does not affect the lesion position, such as rectum, cervix, prostate and lymph nodes, are usually treated in FB. A secondary parameter that is taken into account to select the most appropriate delivery technique, but not less important, is the patient's compliance. In particular, for the BH delivery technique, is extremely important to accurately examine and evaluate the simulation MRI images dataset: if the patient is not suitable to proceed a BH treatment, another delivery technique (FB) is considered, as well as to move the patient to a standard linear accelerator. Despite the advantages offered by real-time motion monitoring, there still are some concerns on the fact that the beam delivery time can be greatly prolonged in the case of inspiratory BH gating treatment.
The efficiency of RT delivery on mobile targets is extremely important in this context, as the average beam delivery time for respiratory-gated irradiation can be two to five times longer than the free breathing one for equivalent fractionation.
As treatment room throughput and efficiency management is also crucial for this innovative technology, some experiences have been published to evaluate treatment process time, 9,10 considering the system log data. 11 Recording these parameters has a huge impact also for optimizing the daily treatment room schedule into specific time slots. 9,10 To the best of our knowledge, treatment process time efficiency for the MRgRT system has not yet been analyzed.
This study aims to evaluate treatment process time efficiency based on our clinical experience, and to propose a new treatment room management approach in the case of BHI or FB conditions.

2.A | Patients database
A sample of consecutive patients treated in our institution with the MRIdian MR-Linac (ViewRay Inc, Mountain View, CA, USA) was considered for this analysis. Patients have been grouped based on of the disease site: lung, liver, pancreas, adrenal gland, lymph node, rectum, cervix and prostate. Both BHI and FB patients were considered for this analysis.

2.B | Treatment planning
All patients included in the study underwent a simulation MRI (0.35 T) and CT scan (GE, Optima CT580 W, HiSpeed DX/I Spiral), acquired sequentially on the same day. Simulation MRI scan is performed with a TRUFI sequence with a steady state precession sequence, image resolution of 1.5 × 1.5 × 1.5 mm 3 and acquisition time of 25 or 175 s. Simulation CT scan, with a slice thickness of 2.5 mm, is acquired in the same position and with the same immobilization and positioning system used in the simulation MRI. During the MRI simulation, one or more sagittal cine MRI (4 frames/s) is also acquired to further define if the treatment will be performed in BH of FB delivery technique, considering both clinical/dosimetric requirements and patient's compliance. Intensity modulated radiation therapy (IMRT) step and shoot treatment plans were calculated using the MRIdian treatment planning system (TPS).
For each treatment plan, the dose was prescribed according to the planning target volume (PTV) and normalized to the 50% of the PTV (in case of homogeneous dose prescription) or to the 80% isodose line (in case of inhomogeneous dose prescription).
MRI-Linac can be used either for standard treatment delivery or in adaptive online modality. Adaptive online modality consists in PLACIDI ET AL.
adapting the treatment plan every day on the basis of the inter-fraction changes in internal anatomy, ensuring the best dose distribution. 12 For each patient treated with online adaptive modality, the evaluation of the daily anatomy could have led to a newly optimized plan, which is re-optimized just some minutes before fraction delivery. For the patients not candidate for online adaptive delivery, offline replanning was performed, if necessary.
2.C | Real-time gating magnetic resonance-guided radiotherapy system Two key parameters have been defined for the sagittal cine MRI (4 frames/s) image acquisition prior to treatment delivery start: 1. Gating boundary, defined as a margin from CTV which will take into account target intra-fraction maximum allowed motion. It depends on anatomical site and patients' characteristics but is generally set to 3-5 mm.
2. Maximum percentage of gating target volume (ROI%), defined as the maximum allowed percentage of the target volume that should be outside the defined boundary to stop beam delivery.
When this threshold value is exceeded, the beam is automatically interrupted. Also ROI% appears to be strongly dependent on the anatomical site and patients' characteristics but it is generally set based on the target structure volumes (V < 8cc, 5 cc < V<20cc, V> 20 cc), respectively to 3-5-8%.  were also considered for this study. Machine and TPS log data have been used in order to extract the aforementioned variables.
In particular, beam on delivery time is defined as: where X is the number of fields per session, V is the CTV in cc and R is the gating function (R = 1 with active gating and R = 0 without).
T Bod (X,V,R) is the actual time of beam delivery, nominally counted when MU are delivered.
The mechanical treatment time (T G Þ is defined for gantry rotation and MLC configuration respectively as: In the delivery treatment log data, actual beam on time T Abo (X,V, R) is recorded as: The actual total treatment process time (T Attp Þ, is defined as: Finally, treatment process time efficiency (defined as treatment efficiency T E (X,V,R)) is expressed as: Real-time sagittal MRI acquisition during treatment delivery (top images is a lung case, bottom images is a liver case): target (red) is irradiated (BEAM ON) only when it is located within the boundary (green). If a defined percentage of the target is outside of the boundary, delivery is automatically interrupted (BEAM OFF).
where T Attp (X,V,R) is the actual total treatment process time (for each single fraction) and T TPSttp (X,V,R) is the predicted treatment process time computed by the TPS during the plan optimization.
Negative values of T E (X,V,R) mean that the TPS has overestimated the treatment process time needed. T E (X,V,R), defined as percentage, has been analyzed for each single fraction of all the patients enrolled in the study and displayed in a Whisker plot. 13 Welch two sample t-test has been performed to evaluate the difference between the considered patients' groups (BHI vs FB).  To summarize, this study has three principal aims:

2.E | Treatment slot time
• treatment process time efficiencycomparison of patients treated in BHI and FB conditions, introducing an efficiency score (T E ) that essentially indicates how close the treatment time provided by the TPS (total, mechanical and beam on) is to the actual delivery one; • treatment slot time suitabilityquantitative assessment of the duration of treatment slots foreseen for BHI patients, in order to optimize the daily treatment room scheduling and the total number of procedures; • clinical, planning and delivery parameters correlationinvestigating possible correlations between treatment process efficiency and clinical/planning/delivery parameters

| RESULTS
The first 100 patients treated in our institution between June and November 2019, corresponding to 919 treatment fractions, were considered for this analysis.

3.A | Treatment process time efficiency
Results (mean and standard deviation, SD) of the actual beam on time (T Abo (X,V,R)), actual total treatment process time (T Attp (X,V,R)), | 73 predicted treatment process time as computed by the TPS (T TPSttp (X, V,R)) and treatment process time efficiency (T E (X,V,R)) for patients treated with (BHI) and without (FB) active respiratory control, are reported in Table 2.
Treatment efficiency (T E (X,V,R)) results are also depicted in term of Whisker plot in Figs Overall T E (X,V,R) results of the two evaluated groups of patients are further summarized in Fig. 4, showing a statistically significant difference (P < 0.001) with a mean value of 42.4% and −0-5% for the BHI and FB group, respectively.

3.B | Treatment slot time
The time required for the following PTP was evaluated (mean (minmax)) in order to investigate the appropriateness of the foreseen treatment slot time occupancy: • Patient ready for positioning: 5 min (3.5-6.5 min) • Patient positioning: 4 min (3-5 min) • Positioning correction: 4 min (1.5-7.5 min) • Adaptive process: 30 min (15-60 min) The average pretreatment time is 13 min for nonadaptive patients (14 and 12 for SBRT and long course treatment respectively), 43 min for adaptive ones. Table 2 describes actual total treatment process times (T Attp (X,V,R), while the actual treatment slot time occupancy (T Attp (X,V,R) + PTP) is described in Table 3.   can prevent bulk motion of the patient on the treatment couch or detect any possible changes in the target position due to a different bladder filling.

3.C | Clinical, planning and delivering parameters
The use of T E (X,V,R) can be a valid indicator to be used in daily clinical practice for gating MRI-Linac activity optimization. Not surprisingly, as also shown in Fig. 4, a statistically significant difference (P < 0.001) was found between the T E (X,V,R) of BHI and FB groups, with a mean value of 42.4% and −0.5% respectively. As reported in the results section, T E (X,V,R) also highlighted relevant intra-patient and intra-site differences among the BHI and FB groups.
Based on the results listed in Table 2  The results reported in Table 3  Clinical, planning and delivery parameters were also evaluated, as listed in Table 4, in order to better understand if T E (X,V,R) could be further optimized with these parameters, on the basis of the observed results. No significant dependence has been observed between time and these parameters.
As far as the authors know, this is the first study where treatment process time (T E (X,V,R)) has been quantitatively analyzed using data obtained from patients planning and treatment logs in order to Liu and colleagues claimed that the accuracy of the proposed prediction algorithm was sufficient to support patient treatment appointment scheduling and was a reliable indicator for treatment plan complexity. 18 Nevertheless, delivery time efficiency due to the gating system activation was not included in the proposed analysis.
As shown in our experience, T E (X,V,R) is a useful parameter able to successfully support the optimization of the treatment room scheduling, integrating gating related variables.
It has to be mentioned that patient compliance is an extremely variable parameter, difficult to quantify and also subject to important inter-fraction T E (X,V,R) variability. In order to evaluate and quantify patient's compliance, several protocols have been included in our clinical practice. Firstly, during the treatment preliminary medical examination, the radiation oncologist verifies the patient's compliance by means of a dedicated questionnaire. In addition to general and clinical information, particular attention is given to MRI safe information, patient compliance and ability to breathing managing during treatment. 19 Once the patient is eligible for MRgRT, a training session to verify and optimize the patient compliance and his/her ability in breathing management is performed during simulation.
As shown in Fig. S1, an evident and generalized learning curve is not feasible. BHI liver patients are probably the most stable group showing a minor inter-fraction variation of the T E (X,V,R). Further analysis, probably with higher statistics, could definitely better investigate this issue.
One potential method to limit the intra-fraction variability and reproducibility in breath hold treatments can be represented by the use of abdominal compression that, on the other side, maybe not compatible with MR-Linac bore dimensions and coils placement. Even if the most common clinical indications are represented in the enrolled population, a selection bias can be recognized in our study, as patients affected by comorbidities that could affect their breathing cycle length and performance have been discarded due to the higher general fitness level required for MRI compatibility. Furthermore, diseases not primarily addressable to MRI-Linac treatments have been directly excluded (i.e. H&N and brain), even if the use of gating systems in these anatomical sites is negligible.
Another limitation is the difference in treatment sites between the FB and BHI groups: nevertheless, it would have not been appropriate to use a BH technique for some of the FB sites or conversely.
This is also because this retrospective analysis has been performed on each of the clinical delivered fractions for both FB and BHI patients and therefore is not feasible to compare FB and BHI approaches for the same treatment site, even prospectively since it would lead a suboptimal treatment for the patients.
Another limitation of this study is that patient repositioning time, in case of bulk motion or treatment position variation during delivery, has not been recorded. This circumstance may represent an unexpected source of treatment time extension that could impact on the time slot distribution throughout daily machine activity.
Finally, further technical and technological developments, such as novel tracking algorithms and higher frame per second real-time cine MR imaging, could potentially affect T E (X,V,R) and this study can represent a robust baseline to compare future development.

CONFLI CT OF INTEREST
The authors declare no conflict of interest.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of the article. Inter-fraction variability of treatment time efficiency for the group of patients treated with gating system (breath hold inspiration). Only four BHI sites have been considered (liver, pancreas, lymph node and lung) and only for patients with more than two treatment fractions.