Effect of respiration-induced motion on a three-dimensional magnetic resonance imaging-based adaptive radiotherapy workflow in a 1.5T magnetic resonance linear accelerator

Introduction

Magnetic resonance imaging (MRI) plays a crucial role in external beam radiotherapy (RT) due to its superior soft-tissue contrast, which enables precise tumor delineation and enhances adaptive treatment planning (1-3). This is particularly crucial for tumors near critical structures, where exact alignment can significantly impact outcomes (4,5). Integrating MRI with linear accelerators (Linac) into MR-Linac systems allows for real-time imaging that facilitates adaptive planning and precise treatment adjustments (6,7). In MR-guided RT, the adapt-to-shape (ATS) workflow, which is uniquely implemented on the Elekta Unity MR-Linac system is essential for cases involving significant anatomical changes that require the redelineation of the target and organs at risk (OARs) to ensure accurate dose distribution (8-10).

However, respiratory-induced motion presents significant challenges in MR-guided RT, primarily the motion artifacts that degrade MRI image quality (11-14). These artifacts include blurring, ghosting, signal loss, and undesired strong signals. Blurring and ghosting primarily stem from signal readout processes, where the movement of structures during acquisition causes a loss of clarity or the generation of multiple replicas (15,16). These may result in inaccurate delineation and dose misalignment, potentially compromising tumor control and increasing the risk of toxicity to adjacent healthy tissues (17).

Conventional motion management techniques, such as four-dimensional computed tomography (4D CT), are widely used in RT to assess and incorporate respiratory motion effects (18-20). Although MRI offers superior soft-tissue contrast, it is inherently more sensitive to motion artifacts compared to CT, complicating target visualization during respiratory motion (21). Advanced imaging techniques, including 4D MRI (22,23) and cine MRI (24,25), have shown potential in managing motion by capturing respiratory patterns. However, these methods are not yet fully standardized in clinical practice and face limitations in capturing complete breathing cycles. With the continued advancement of MR-Linac systems, there is a growing need to systematically evaluate the effect of respiratory motion on ATS workflows, particularly in achieving the visual and spatial accuracy required for optimal dose distribution.

Understanding and quantifying the effect of respiratory motion in MR-guided RT is clinically significant, as it directly influences delineation precision and dosimetric accuracy (26). Motion parameters, such as amplitude, frequency, and variability, can alter the apparent size and position of tumors on MRI, potentially causing discrepancies in dose distribution if not adequately addressed during treatment (17). Motion artifacts may distort the perceived tumor motion range, leading to inaccurate target delineation that either underestimates or overestimates actual displacement, potentially resulting in tumor underdosing or excessive radiation to adjacent critical structures. These challenges underscore the need for the development of adaptive workflows tailored to individual patient's respiratory patterns to minimize uncertainty and improve treatment outcomes.

This study simulated tumor respiratory-induced linear translational motion using an MR-compatible motion phantom with respiratory motion curves from six patients, each generating eight distinct respiratory patterns, resulting in 48 simulated scenarios. The impact of respiratory-induced translational motion on the MR-guided ATS workflow was quantitatively assessed through key delineation and dosimetric metrics.

Methods

Motion phantom and patterns

The Zeus MR-compatible motion phantom (Sun Nuclear, Colbourne, FL, USA) was used to simulate respiratory-induced tumor motion (Figure 1). The phantom comprises anatomically shaped, static compartments representing key thoracic and abdominal organs, namely the lungs, liver, kidneys, and spine, each filled with contrast-providing gels suitable for both computed tomography (CT) and MRI imaging. The moving component, designed to replicate a tumor, features an irregular shape embedded within the structure. It is capable of independent, programmable linear movements with an accuracy of ±0.2 mm and rotational movements with an accuracy of ±0.25°. The piezoelectric motors enable motion in the superior-inferior (SI), anterior-posterior (AP), and left-right (LR) directions.

Figure 1 Zeus magnetic resonance motion phantom. (A) Photograph of the Zeus MR motion phantom and dynamic motion controller. (B) Three-dimensional reconstruction of simulated tumor (orange), lungs (green), liver (red), kidney (blue), and spine (purple).

To replicate patient-specific respiratory-induced translational motion, respiratory motion curves were retrospectively collected from six patients via a real-time position management system (Varian Medical Systems, Palo Alto, CA, USA) during 4D CT scans, with a sampling of 0.04 seconds. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the ethics committee of the West China Hospital (No. 20232269) and individual consent for this retrospective analysis was waived. These curves, reflecting each patient’s unique respiratory irregularities, were imported into CIRS motion control software (Sun Nuclear) to precisely guide the phantom’s movements. Figure 2 presents these respiratory curves, illustrating the diversity of respiratory patterns. To simulate a wide range of clinical scenarios, the maximum amplitude and frequency in the SI direction were adjusted for each curve, generating eight distinct respiratory motion combinations [see Supplementary file (Appendix 1)]. These combinations, detailed in Table 1, represented respiratory-induced target motions ranging from shallow to deep breathing patterns and were used to evaluate target motion and adaptive RT accuracy under varying respiratory conditions.

Figure 2 Patients’ respiratory curves with amplitude normalized to 10 mm and frequency normalized to 15 min⁻¹.

Treatment planning

For pretreatment imaging, CT and MR scans were acquired with the phantom in a stationary state via a Revolution ES CT scanner (GE HealthCare, Chicago, IL, USA) and the Unity MR-linac (Elekta AB, Stockholm, Sweden), respectively. For the CT simulation (see Figure 3A), the tube voltage was set to 120 kV, with a slice thickness of 0.625 mm. The tumor target and adjacent structures, namely the lungs, liver, kidneys, and spinal cord, were delineated with the Monaco treatment planning system (v. 5.40.04; Elekta AB). A reference treatment plan (see Figure 3B) was generated with a 9-field beam arrangement at angles of 20°, 60°, 100°, 140°, 181°, 220°, 260°, 300°, and 340°. The prescribed dose was 4,500 cGy in 15 fractions, with optimization parameters determined by a senior physicist [see Supplementary file (Appendix 2)]. Dose calculations involved a grid resolution of 2 mm and a statistical uncertainty of 1%. After optimization, the dose received by 95% of the volume (D₉₅) of the gross tumor volume (GTV) reached 100% of the prescribed dose, ensuring clinical compliance with target conformity and coverage standards.

Figure 3 The workflow for the simulation and planning of respiratory-induced tumor linear translational motion with the Zeus motion phantom. The process includes CT simulation (A), treatment planning (B), MR-based digital reference model (C), MR scanning (D), and an ATS workflow (E,F). ATS, adapt-to-shape; CT, computed tomography; ITV, internal target volume; ITVref, reference internal target volume; MR, magnetic resonance.

MR-based digital reference model

The MR images of the Zeus phantom were acquired in its stationary state through use of standard spin-spin relaxation time-weighted (T2W) 3D MR sequence settings for abdominal imaging, which served as a baseline reference. The default scanning parameters were as follows: field of view (FOV) =360 (AP) ×447 (RL) ×300 mm (SI), matrix size =180 (AP) ×223 (RL) ×250 (SI), reconstructed voxel size 2×2×2.4 mm³, repetition time (TR) =2,100 ms, time to echo (TE) =212 ms, turbo spin echo (TSE) factor =134, flip angle (FA) =90º, and number of signal average (NSA) =2. The total scan duration was 200 seconds. These images were used to establish the initial positioning with the target at a displacement value of 0 mm and to delineate the structure of the GTV and surrounding anatomy in the absence of motion.

Based on predefined respiratory motion curves, both reference internal target volume (ITV_ref) structures and average intensity projection (AIP) images were generated to capture the tumor’s full motion throughout the respiratory cycle (see Figure 3C). MR images of the Zeus phantom in its stationary state were used as a baseline. According to the displacement values at each sampling time point specified by the respiratory motion curves, the voxel signal values in the region of moving component were reassigned via linear interpolation to generate the MR image set and the corresponding GTV structure based on these images. The AIP images were obtained by averaging the voxel values of MR images at each sampling time point, while the reference ITV (ITV_ref) structures were obtained by combining the GTVs from each sampling time point. To validate the accuracy of the MR digital phantom, rigid registration was performed between the generated MR images and corresponding MR images of the phantom in its stationary state, with the same displacement values being used to ensure that the simulated respiratory motion accurately reflected the actual positional changes [see Supplementary file (Appendix 3)].

ATS workflow

For each respiratory motion pattern, online T2W MRI scans were acquired under default scanning parameters (see Figure 3D), resulting in a total of 48 MR image sets (6 patients × 8 motion patterns). Under the same reference plan, the ATS workflow was executed for each motion pattern, with ITV structures delineated by an experienced clinician (see Figure 3E). Additionally, 48 AIP image sets and ITV_ref structures were processed through the ATS workflow as reference comparisons (see Figure 3F). An average density derived from the GTV was assigned to each full ITV and was used for planning. These ITV and ITV_ref structures were designated as the optimization target, with 95% of each ITV receiving 100% of the prescribed dose coverage. For dose assessment, the ITV_ref structures were overlaid onto the corresponding MRI sets. In total, 96 adaptive plans (48 ATS plans and 48 reference ATS plans) were generated across all motion patterns to evaluate the accuracy of the ATS workflow in addressing respiratory-induced motion effects on delineation and dosimetry.

Evaluation metrics

The ITV and ITV_ref for each respiratory motion pattern were evaluated and compared using 3D-Slicer software (www.slicer.org). The following metrics were employed to assess the accuracy and consistency of the target volumes:

The Dice similarity coefficient (DSC) was applied to quantify the overlap between two volumes, with a value of 1 indicating perfect agreement and 0 indicating no overlap. It is calculated as follows:

DSC=2×|A∩B||A|+|B|[1]

where A and B are the ITV and ITV_ref volumes, respectively.

The Hausdorff distance (HD) was used to measure the maximum distance between the surface points of the ITV and ITV_ref, reflecting the largest spatial discrepancy between the two volumes. It is calculated as follows:

HD(A,B)=max{supa∈Ainfb∈Bd(a,b),supb∈Binfa∈Ad(a,b)}[2]

where d (a,b) represents the Euclidean distance between points a and b.

The mean surface distance (MSD) is the average distance between corresponding surface points of the ITV and ITV_ref, offering an overall measure of surface agreement. It is calculated as follows:

MSD(A,B)=1|SA|+|SB|(∑a∈SAinfb∈SBd(a,b)+∑b∈SBinfa∈SAd(a,b))[3]

where S_A and S_B are the surface points of the ITV and ITV_ref, respectively.

The volume difference (VD) metric was used to quantify the difference in volume between the ITV and ITV_ref, indicating the overall discrepancy in target size. It is expressed as follows:

VD=|VITV−VITVref|VITVref[4]

where V_ITV and V_ITVref are the volumes of the ITV and ITV_ref, respectively.

The craniocaudal maximum displacement error (CCMDE) was applied to measure the maximum error in the SI direction, capturing the greatest discrepancy in the SI alignment between the ITV and ITV_ref. It is calculated as the absolute difference between the maximum extents of ITV and ITV_ref along the craniocaudal axis.

Additionally, the clinical practicability of ITV delineation was evaluated through an analysis of the accuracy of the dose distribution in the ATS plan design. To assess dose conformity to the ITV, the dose volume histogram (DVH) parameters, such as the dose received by 98% of the volume (D₉₈), D₉₅, mean dose delivered to the entire volume (D_mean), volume receiving 100% of the prescribed dose (V₁₀₀), and volume receiving 95% of the prescribed dose (V₉₅) of ITV_ref from the ATS plan, were subtracted from the corresponding parameters of ITV_ref from the reference ATS plan. The differences in the DVH parameter provided a comprehensive measure of plan quality under varying respiratory motion patterns.

Data analysis

All data were analyzed with SPSS software version 26.0 (IBM Corp., Armonk, NY, USA). The Friedman test was employed to evaluate the effect of different respiratory motion patterns on delineation accuracy and dosimetric outcomes, with statistical significance set at P<0.05. For metrics with significant differences, Spearman rank correlation was used to analyze the relationship between respiratory motion characteristics and outcome metrics. The correlations between respiratory parameters and key metrics were visualized via scatter plots with corresponding trend lines.

Results

Summary of reference ITV volume and plan dosimetry

The V_ITVref values and the dose parameters from the reference ATS adaptive plan are shown in Tables 2,3. The V_ITVref values varied significantly with the respiratory amplitude, ranging from 23.36 to 49.06 cm³. Correspondingly, the range of SI motion increased from 46.23 to 70.07 mm (Table 2). Despite these variations, the key ITV_ref dose parameters (D₉₈, D₉₅, D_mean, V₁₀₀, and V₉₅) remained within clinical standards across different amplitudes. For instance, D₉₅ consistently remained at or near the prescribed dose of 4,500 cGy.

Delineation impact of respiratory translational motion

The delineation accuracy metrics showed significant variations with respiratory amplitude, as presented in the Table 4. The DSC exhibited a marked decline, decreasing from 0.96 at 2.50 mm amplitude to 0.83 at 12.50 mm amplitude (P<0.001), indicating reduced overlap between ITV and ITV_ref as the amplitude increased. The HD consistently increased from 2.62 to 6.32 mm, while the MSD rose from 0.08 to 0.64 mm (both P values <0.001). Similarly, the VD increased significantly from 7.97% to 26.80% (P<0.001), highlighting substantial volumetric discrepancies at higher amplitudes. The CCMDE followed a similar pattern, increasing from 2.37 to 13.33 mm (P<0.001).

Respiratory frequency exhibited minimal influence on delineation metrics, as shown in Table 5. Across frequencies ranging from 10 to 20 min⁻¹, the DSC had an offset range of 0.01, indicating negligible variability in the overlap between ITV and ITV_ref. Similarly, the HD had an offset range of 0.10 mm, while the MSD demonstrated a variation of 0.05 mm. For volumetric discrepancies, the VD varied by 1.53%, and the CCMDE showed an offset of 1.04 mm. These variations across delineation metrics suggest that respiratory frequency had a limited impact on spatial and volumetric accuracy, particularly when compared to the substantial effects observed with respiratory amplitude. No significant differences were observed in the delineation accuracy metrics across various respiratory frequencies (all P values >0.05).

Figure 4 presents the coronal MR images of the phantom using the eight motion patterns for the motion curves of patient 1. The pink contours represent the ITV_ref, while the yellow contours represent the delineated ITV. With increasing respiratory amplitude, there was a noticeable divergence between the ITV_ref and delineated ITV, indicating reduced spatial accuracy in the delineation process. In contrast, variations in respiratory frequency had a less pronounced effect, as the alignment between the ITV_ref and the delineated ITV remained relatively stable across different frequencies.

Figure 4 Coronal MR images of the phantom under the eight respiratory motion patterns derived from the breathing curve of patient 1. The pink contours represent the ITV_ref, while the yellow contours represent the delineated ITV. (A-H) Different combinations of respiratory amplitudes and frequencies. MR, magnetic resonance; ITV_ref, reference internal target volume.

Dosimetric impact of respiratory translational motion

The differences in dosimetric parameters showed significant variations with respiratory amplitude, as presented in Table 6. Increasing respiratory amplitude resulted in pronounced discrepancies in ITV_ref dose metrics. The median difference in D₉₅ increase from 17.70 cGy at 2.50 mm amplitude to 481.55 cGy at 12.50 mm amplitude (P<0.001), highlighting a clear reduction in target dose coverage at higher motion amplitudes. Similar trends were observed for V_100, with negative differences rising from 3.42% to 20.13% (P<0.001), indicating a substantial decrease in the volume of the ITV_ref receiving the full prescribed dose.

In contrast to amplitude, respiratory frequency had a limited impact on dosimetric differences, as shown in Table 7. Across frequencies ranging from 10 to 20 min⁻¹, the offset range of ∆D₉₅ was 95.15 cGy, with median values varying from 217.10 to 312.25 cGy (P>0.05). Similarly, the offset range of ∆V₁₀₀ was 2.47%, with values fluctuating between 14.91% and 17.38% (P>0.05), while ∆V₉₅ exhibited an offset range of 2.19%, spanning from 4.84% to 7.03% (P>0.05). These minor variations highlight the minimal impact of frequency changes on target dose coverage metrics.

Relationship between respiratory and key metrics

Based on the Friedman test results from “Delineation impact of respiratory translational motion” and “Dosimetric impact of respiratory translational motion” sections, Spearman rank correlation was conducted to evaluate the relationship between respiratory amplitude and key delineation and dosimetric metrics. As shown in Figure 5, all evaluated metrics exhibited significant correlations with respiratory amplitude.

Figure 5 Relationship between respiratory amplitude and the key metrics of delineation and dosimetry. Scatterplots illustrate the impact of maximum respiratory amplitude on (A-E) delineation metrics and (G-K) dosimetric metrics. Linear trends are shown with corresponding 95% confidence bands for each metric. (F) The legend for the delineation metrics. (L) The legend for the dosimetric metrics. Spearman correlation coefficients (ρ) and slopes are annotated for each graph. CCMDE, craniocaudal maximum displacement error (mm); DSC, Dice similarity coefficient; HD, Hausdorff distance (mm); MSD, mean surface distance (mm); VD, volume difference (%); ∆D₉₈, difference in dose covering 98% of the ITV_ref (cGy); ∆D₉₅, difference in dose covering 95% of the ITV_ref (cGy); ∆D_mean, difference in mean dose within the ITV_ref (cGy); ∆V₁₀₀, difference in volume of ITV_ref receiving 100% prescribed dose (%); ∆V₉₅, difference in volume of ITV_ref receiving 95% prescribed dose (%).

For delineation metrics, the DSC displayed a negative correlation with amplitude, with a slope of –0.012 mm⁻¹ (ρ=–0.975), indicating a steady decline in overlap between ITV and ITV_ref as amplitude increased (Figure 5A). The HD and MSD showed positive correlations, with slopes of 0.467 mm/mm (ρ=0.868) and 0.062 mm/mm (ρ=0.964), respectively, reflecting greater geometric discrepancies at higher amplitudes (Figure 5B,5C). Similarly, volumetric inaccuracies increased, as evidenced by the VD and CCMDE, which had slopes of 1.896 %/mm (ρ=0.972) and 1.148 mm/mm (ρ=0.959), respectively (Figure 5D,5E).

Regarding the dosimetric metrics, the ∆D₉₈, ∆D₉₅, and ∆D_mean values exhibited slopes of 124.0 cGy/mm (ρ=0.945), 55.95 cGy/mm (ρ=0.950), and 10.97 cGy/mm (ρ=0.900), respectively, reflecting reduced dose coverage with increasing respiratory amplitude (Figure 5G-5I). Similarly, target coverage metrics showed significant reductions, with ∆V₁₀₀ and ∆V₉₅ demonstrating slopes of 1.761 %/mm (ρ=0.959) and 1.097 %/mm (ρ=0.937), respectively, indicating a decline in the percentage of the ITV covered by the prescribed dose as the amplitude increased (Figure 5J,5K).

Discussion

This study systematically evaluated the impact of respiratory-induced linear translational motion on delineation accuracy and dosimetric outcomes within the ATS workflow of MR-guided RT. MR-guided RT offers significant advantages through its superior soft-tissue contrast, enabling precise target delineation and adaptive treatment planning, particularly for tumors in anatomically complex regions. However, respiratory motion remains a critical challenge, introducing motion artifacts that compromise both geometric precision and dose distribution accuracy. A systematic analysis of respiratory motion-induced uncertainties is essential to identifying sources of error and developing effective strategies for optimizing adaptive treatment workflows, ultimately enhancing clinical outcomes.

The statistical analysis of delineation and dosimetric metrics revealed that respiratory amplitude significantly influences both geometric and dosimetric accuracy within 3D Cartesian MR-guided adaptive RT workflows. The impact of respiratory amplitude observed in this study aligns with the findings of Bertelsen et al., who identified amplitude as the dominant factor affecting geometric errors in MR imaging (26). Their research demonstrated that amplitude-driven artifacts, such as blurring and positional shifts, vary depending on respiratory translational motion patterns, namely, symmetric or asymmetric breathing cycles. Building on this understanding, our study integrated both geometric and dosimetric evaluations, offering a more comprehensive perspective on motion-related challenges in MR-guided RT. Notably, a near-linear relationship was observed between amplitude and delineation metrics, such as the DSC and HD. Larger amplitudes consistently resulted in poorer delineation accuracy and reduced target overlap. These geometric discrepancies further affected dosimetric outcomes, leading to decreased ITV dose coverage. In contrast, respiratory frequency exhibited little impact on both delineation and dosimetric metrics. In our study, the MR sequence employed a TSE factor of 134, with each TR (2,100 ms) acquiring 134 k-space lines, corresponding to an acquisition time of approximately 15.7 ms per line. During each respiratory cycle (3–6 seconds), a large amount of crucial low-frequency data are rapidly captured, and the overall 200 seconds scan spans multiple respiratory cycles. Temporal averaging during image reconstruction reduce the effect of respiratory frequency variations on image quality (27,28).

The geometric and dosimetric inaccuracies observed in our study can largely be attributed to the prolonged acquisition time of 3D Cartesian MRI, which effectively averages tumor motion over the respiratory cycle. This results in images visually similar to AIP images, which are effective for visualizing the extent of tumor motion but are not recommended for direct use in ITV delineation due to their inability to capture precise spatial boundaries. These findings align with those of Borm et al., who demonstrated that motion-averaged CT images, such as AIP derived from 4D CT, provide a useful representation of motion extent but lead to blurred and geometrically inaccurate target depictions (29). Although Borm et al.’s study was based on 4D CT, the parallels underscore a shared limitation of motion-averaged imaging techniques in the accurate delineation of dynamic targets. This highlights the importance of employing more advanced imaging approaches, such as 4D MRI or real-time gated imaging, to achieve precise ITV delineation in MR-guided RT.

These observations highlight the necessity of robust motion management strategies, particularly for cases of thoracic and abdominal tumors in which respiratory motion is most pronounced. By incorporating patient-specific respiratory waveforms, this study effectively replicated realistic clinical scenarios, offering insights into the complex interplay between respiratory motion characteristics and treatment accuracy. The findings demonstrate that respiratory amplitude has a far greater impact on delineation and dosimetry than does frequency, underscoring the importance of strategies aimed at minimizing amplitude. Techniques such as abdominal compression (30) and deep-inspiration breath hold (31) are effective in reducing respiratory amplitude, thereby mitigating motion-induced delineation and dosimetric errors. Furthermore, advanced imaging technologies offer opportunities to improve currently used clinical workflows, as precise tumor imaging and accurate quantification of tumor motion are critical. Methods for mitigating respiratory motion artifacts in MR imaging include motion prevention (breath-hold) (32), artifact reduction (fast imaging) (33), motion-insensitive sequences such as radial sequences (34), and motion correction (MR navigators) (11,35,36). These techniques reduce or eliminate motion artifacts, thus decreasing the difficulty in accurately delineating the target. Based on clear tumor imaging, respiratory motion characteristics for each patient and each fraction can be obtained using 2D MR cine and 4D MRI, followed by patient-specific ITV expansion to ensure adequate dose coverage of the target. Integrating these strategies into clinical workflows can significantly enhance the accuracy and efficacy of MR-guided RT, ultimately improving patient outcomes.

Although this study provides useful insights into the impact of respiratory-induced translational motion on MR-guided RT, certain limitations should be acknowledged. First, the study relied on a motion phantom to replicate patient-specific respiratory-induced translational motion, and despite being driven by patient-derived respiratory waveforms, the phantom does not fully capture the complexity of real-world clinical scenarios. Factors such as tumor deformation, organ elasticity, rotational motion, and the complex interactions of respiratory motion with surrounding anatomical structures were not considered. Second, the use of 3D Cartesian MRI sequences introduces motion-averaging artifacts due to prolonged acquisition times. These sequences generate motion-averaged images, visually similar to those of AIP, which limits their suitability for precise ITV delineation. Moreover imaging modalities, such as 4D MRI or cine MRI, were not used in this study. These techniques, however, have the potential to provide dynamic, time-resolved imaging, offering improved spatial and temporal accuracy for tracking respiratory-induced motion.

Building upon the limitations identified in this study, several promising directions for future research and clinical applications can be proposed. The integration of advanced imaging techniques, such as 4D MRI or cine MRI, into MR-guided RT workflows represents a critical step forward. These technologies provide dynamic, time-resolved imaging, enabling more accurate representations of respiratory-induced motion and addressing the shortcomings of motion-averaged 3D Cartesian MRI. Including these modalities into adaptive workflows could significantly improve the precision of ITV delineation and dose delivery. To further optimize motion management strategies, future research should evaluate the effectiveness of techniques such as gated RT, breath-hold approaches, and abdominal compression within MR-guided RT. Such studies would provide evidence-based recommendations, ultimately enhancing the accuracy and efficacy of MR-guided RT in clinical practice.

Conclusions

This study evaluated the impact of respiratory-induced linear translational motion on the delineation accuracy and dosimetric outcomes within the ATS workflow of MR-guided RT through the use of a motion phantom and patient-specific respiratory waveforms. Larger respiratory amplitudes significantly compromised geometric and volumetric accuracy, as evidenced by declines in delineation accuracy and dose coverage metrics. In contrast, respiratory frequency had minimal effects on these metrics. This study also identified certain limitations of 3D Cartesian MRI, as motion-averaged artifacts hindered precise ITV delineation, emphasizing the need for advanced imaging modalities for achieving greater spatial and temporal accuracy.

Acknowledgments

None.

Footnote

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2024-2866/dss

Funding: This work was supported by research grants from National Key R&D Program of China (No. 2022YFC2404605) and National Natural Science Foundation of China (Nos. 12475348 and 12205209).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2024-2866/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the ethics committee of the West China Hospital (No. 20232269) and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Yuan J, Poon DMC, Lo G, Wong OL, Cheung KY, Yu SK. A narrative review of MRI acquisition for MR-guided-radiotherapy in prostate cancer. Quant Imaging Med Surg 2022;12:1585-607. [Crossref] [PubMed]
2. De Pietro S, Di Martino G, Caroprese M, Barillaro A, Cocozza S, Pacelli R, Cuocolo R, Ugga L, Briganti F, Brunetti A, Conson M, Elefante A. The role of MRI in radiotherapy planning: a narrative review "from head to toe". Insights Imaging 2024;15:255. [Crossref] [PubMed]
3. Paganelli C, Whelan B, Peroni M, Summers P, Fast M, van de Lindt T, McClelland J, Eiben B, Keall P, Lomax T, Riboldi M, Baroni G. MRI-guidance for motion management in external beam radiotherapy: current status and future challenges. Phys Med Biol 2018;63:22TR03. [Crossref] [PubMed]
4. Liu X, Li Z, Yin Y. Clinical application of MR-Linac in tumor radiotherapy: a systematic review. Radiat Oncol 2023;18:52. [Crossref] [PubMed]
5. Wang H, Yang J, Lee A, Phan J, Lim TY, Fuller CD, Han EY, Rhee DJ, Salzillo T, Zhao Y, Chopra N, Pham M, Castillo P, Sobremonte A, Moreno AC, Reddy JP, Rosenthal D, Garden AS, Wang X. MR-guided stereotactic radiation therapy for head and neck cancers. Clin Transl Radiat Oncol 2024;46:100760. [Crossref] [PubMed]
6. Randall JW, Rammohan N, Das IJ, Yadav P. Towards Accurate and Precise Image-Guided Radiotherapy: Clinical Applications of the MR-Linac. J Clin Med 2022;11:4044. [Crossref] [PubMed]
7. Raaymakers BW, Jürgenliemk-Schulz IM, Bol GH, Glitzner M, Kotte ANTJ, van Asselen B, et al. First patients treated with a 1.5 T MRI-Linac: clinical proof of concept of a high-precision, high-field MRI guided radiotherapy treatment. Phys Med Biol 2017;62:L41-50. [Crossref] [PubMed]
8. Axford A, Dikaios N, Roberts DA, Clark CH, Evans PM. An end-to-end assessment on the accuracy of adaptive radiotherapy in an MR-linac. Phys Med Biol 2021;66:055021. [Crossref] [PubMed]
9. de Leon J, Crawford D, Moutrie Z, Alvares S, Hogan L, Pagulayan C, Jelen U, Loo C, Aylward JD, Condon K, Dunkerley N, Heinke MY, Sampaio S, Simon K, Twentyman T, Jameson MG. Early experience with MR-guided adaptive radiotherapy using a 1.5 T MR-Linac: First 6 months of operation using adapt to shape workflow. J Med Imaging Radiat Oncol 2022;66:138-45. [Crossref] [PubMed]
10. Gupta A, Dunlop A, Mitchell A, McQuaid D, Nill S, Barnes H, Newbold K, Nutting C, Bhide S, Oelfke U, Harrington KJ, Wong KH. Online adaptive radiotherapy for head and neck cancers on the MR linear Accelerator: Introducing a novel modified Adapt-to-Shape approach. Clin Transl Radiat Oncol 2022;32:48-51. [Crossref] [PubMed]
11. Zaitsev M, Maclaren J, Herbst M. Motion artifacts in MRI: A complex problem with many partial solutions. J Magn Reson Imaging 2015;42:887-901. [Crossref] [PubMed]
12. Walker A, Liney G, Holloway L, Dowling J, Rivest-Henault D, Metcalfe P. Continuous table acquisition MRI for radiotherapy treatment planning: distortion assessment with a new extended 3D volumetric phantom. Med Phys 2015;42:1982-91. [Crossref] [PubMed]
13. Torfeh T, Hammoud R, El Kaissi T, McGarry M, Aouadi S, Fayad H, Al-Hammadi N. Geometric accuracy of the MR imaging techniques in the presence of motion. J Appl Clin Med Phys 2018;19:168-75. [Crossref] [PubMed]
14. Yue Y, Fan Z, Yang W, Pang J, Deng Z, McKenzie E, Tuli R, Wallace R, Li D, Fraass B. Geometric validation of self-gating k-space-sorted 4D-MRI vs 4D-CT using a respiratory motion phantom. Med Phys 2015;42:5787-97. [Crossref] [PubMed]
15. Van de Walle R, Lemahieu I, Achten E. Magnetic resonance imaging and the reduction of motion artifacts: review of the principles. Technol Health Care 1997;5:419-35.
16. Wood ML, Henkelman RM. MR image artifacts from periodic motion. Med Phys 1985;12:143-51. [Crossref] [PubMed]
17. Liu X, Li C, Zhu J, Gong G, Sun H, Li X, Sun M, Zhang Z, Li B, Yin Y, Li Z. Technical Note: End-to-end verification of an MR-Linac using a dynamic motion phantom. Med Phys 2021;48:5479-89. [Crossref] [PubMed]
18. Palmer AL, Nash D, Kearton JR, Jafari SM, Muscat S. A multicentre 'end to end' dosimetry audit of motion management (4DCT-defined motion envelope) in radiotherapy. Radiother Oncol 2017;125:453-8. [Crossref] [PubMed]
19. Kwong Y, Mel AO, Wheeler G, Troupis JM. Four-dimensional computed tomography (4DCT): A review of the current status and applications. J Med Imaging Radiat Oncol 2015;59:545-54. [Crossref] [PubMed]
20. Dumas M, Laugeman E, Sevak P, Snyder KC, Mao W, Chetty IJ, Ajlouni M, Wen N. Technical Note: Comparison of the internal target volume (ITV) contours and dose calculations on 4DCT, average CBCT, and 4DCBCT imaging for lung stereotactic body radiation therapy (SBRT). J Appl Clin Med Phys 2020;21:288-94. [Crossref] [PubMed]
21. Lei K, Syed AB, Zhu X, Pauly JM, Vasanawala SS. Artifact- and content-specific quality assessment for MRI with image rulers. Med Image Anal 2022;77:102344. [Crossref] [PubMed]
22. Feng L. Live-view 4D GRASP MRI: A framework for robust real-time respiratory motion tracking with a sub-second imaging latency. Magn Reson Med 2023;90:1053-68. [Crossref] [PubMed]
23. Xiao H, Han X, Zhi S, Wong YL, Liu C, Li W, Liu W, Wang W, Zhang Y, Wu H, Lee HV, Cheung LA, Chang HC, Liao YP, Deng J, Li T, Cai J. Ultra-fast multi-parametric 4D-MRI image reconstruction for real-time applications using a downsampling-invariant deformable registration (D2R) model. Radiother Oncol 2023;189:109948. [Crossref] [PubMed]
24. Westley RL, Alexander SE, Goodwin E, Dunlop A, Nill S, Oelfke U, McNair HA, Tree AC. Magnetic resonance image-guided adaptive radiotherapy enables safe CTV-to-PTV margin reduction in prostate cancer: a cine MRI motion study. Front Oncol 2024;14:1379596. [Crossref] [PubMed]
25. Sui Z, Palaniappan P, Paganelli C, Kurz C, Landry G, Riboldi M. Imaging error reduction in radial cine-MRI with deep learning-based intra-frame motion compensation. Phys Med Biol 2024;
26. Bertelsen A, Bernchou U, Schytte T, Brink C, Mahmood F. The effect of respiration-induced target motion on 3D magnetic resonance images used to guide radiotherapy. Phys Imaging Radiat Oncol 2022;24:167-72. [Crossref] [PubMed]
27. Huang SY, Seethamraju RT, Patel P, Hahn PF, Kirsch JE, Guimaraes AR, Body MR. Imaging: Artifacts, k-Space, and Solutions. Radiographics 2015;35:1439-60. [Crossref] [PubMed]
28. Hanson HM, Eiben B, McClelland JR, van Herk M, Rowland BC. Technical Note: Four-dimensional deformable digital phantom for MRI sequence development. Med Phys 2021;48:5406-13. [Crossref] [PubMed]
29. Borm KJ, Oechsner M, Wiegandt M, Hofmeister A, Combs SE, Duma MN. Moving targets in 4D-CTs versus MIP and AIP: comparison of patients data to phantom data. BMC Cancer 2018;18:760. [Crossref] [PubMed]
30. Grimbergen G, Eijkelenkamp H, Heerkens HD, Raaymakers BW, Intven MPW, Meijer GJ. Dosimetric impact of intrafraction motion under abdominal compression during MR-guided SBRT for (Peri-) pancreatic tumors. Phys Med Biol 2022; [Crossref]
31. Galetto M, Nardini M, Capotosti A, Meffe G, Cusumano D, Boldrini L, Chiloiro G, Romano A, Votta C, Gambacorta MA, Indovina L, Placidi L. Motion and dosimetric criteria for selecting gating technique for apical lung lesions in magnetic resonance guided radiotherapy. Front Oncol 2023;13:1280845. [Crossref] [PubMed]
32. Wang K, Li X, Liu J, Guo X, Li W, Cao X, Yang J, Xue K, Dai Y, Wang X, Qiu J, Qin N. Predicting the image quality of respiratory-gated and breath-hold 3D MRCP from the breathing curve: a prospective study. Eur Radiol 2023;33:4333-43. [Crossref] [PubMed]
33. Johnson CL, Holtrop JL, McGarry MD, Weaver JB, Paulsen KD, Georgiadis JG, Sutton BP. 3D multislab, multishot acquisition for fast, whole-brain MR elastography with high signal-to-noise efficiency. Magn Reson Med 2014;71:477-85. [Crossref] [PubMed]
34. Grimbergen G, Eijkelenkamp H, van Vulpen JK, van de Ven S, Raaymakers BW, Intven MPW, Meijer GJ. Feasibility of online radial magnetic resonance imaging for adaptive radiotherapy of pancreatic tumors. Phys Imaging Radiat Oncol 2023;26:100434. [Crossref] [PubMed]
35. Abdi M, Bilchick KC, Epstein FH. Compensation for respiratory motion-induced signal loss and phase corruption in free-breathing self-navigated cine DENSE using deep learning. Magn Reson Med 2023;89:1975-89. [Crossref] [PubMed]
36. Havsteen I, Ohlhues A, Madsen KH, Nybing JD, Christensen H, Christensen A. Are Movement Artifacts in Magnetic Resonance Imaging a Real Problem?-A Narrative Review. Front Neurol 2017;8:232. [Crossref] [PubMed]