Computed Tomography Simulation for Gamma Knife Radiosurgery: Impact on Target Coverage

Abstract

This study evaluated the effectiveness of using a traditional computed tomography (CT) simulation scan in the Leksell Gamma Knife® Icon™ workflow for stereotactic radiosurgery. We compared two workflows, one involving a traditional CT simulation co-registered with the magnetic resonance imaging (MRI) and cone-beam CT (CBCT) (MRI-CT-CBCT) and another without the CT scan, co-registering the MRI directly with CBCT (MRI-CBCT). The cohort included 50 frame-based (54 targets) and 55 frameless (124 targets) patients treated between August 2020 and December 2022. Target coverage was assessed by the percentage of each target covered by the prescription isodose line. The mean coverage differences for the frame-based and frameless cohorts were −1.20% ± 1.43% and −1.06% ± 1.51%, respectively, with 95% confidence intervals of −1.59, −0.81 and −1.32, −0.79. The MRI-CBCT workflow was deemed noninferior to the MRI-CT-CBCT workflow. Our data indicate that both workflows yield similar results, suggesting that a traditional CT simulation scan is unnecessary.

INTRODUCTION

Leksell Gamma Knife® (LGK) (Elekta AB, Stockholm, Sweden) is a specialized device used for the accurate delivery of therapeutic radiation for brain tumors.[1,2] Numerous design changes have been implemented since the first 1967 LGK model.[3] The latest LGK models, Icon™ and Esprit™, have a cone-beam computed tomography (CBCT) imaging system and a high-definition motion management camera system to assist with image-guided radiosurgery and patient monitoring during treatment.[4,5,6]

Conventionally, on LGK, stereotactic space is defined using the localizer box and imaging (computed tomography [CT], magnetic resonance, or angiography).[7,8,9] However, with the advent of the latest LGK units, stereotactic space can be defined using the onboard CBCT. When defining stereotactic space using the onboard CBCT, other imaging sets are co-registered with a reference CBCT using Leksell GammaPlan’s (LGP’s) built-in co-registration algorithm to establish coordinates. Subsequently, another CBCT is acquired directly before treatment to ensure the stability of the coordinate system. The onboard CBCT has two scanning presets available for imaging: a low dose (CT dose index [CTDI] = 2.5 mGy) and a high quality (CTDI = 6.3 mGy). The high-quality scanning preset is typically used to obtain the reference CBCT, and the low-dose setting is used for subsequent CBCT images.[10,11,12,13]

Co-registration between different imaging modalities used in treatment planning for LGK patients is important for accurate target localization.[14] Previous studies have shown co-registration accuracy between CT and magnetic resonance imaging (MRI) images.[15] The authors reported mean errors of <1 mm in phantom and clinical patient datasets. Previous studies have also shown the image registration accuracy for the LGK Icon™ CBCT workflow.[16,17,18] Chung et al. showed a mean deviation of 0.8 ± 0.3 mm for MRI to CBCT (MRI-CBCT) registration.[16] Ruschin et al. explored the variability of different co-registration workflows for LGK Icon™, MRI to CT, CT to CBCT, and MRI-CBCT. They found that all the workflows were repeatable with <0.2 mm variability.[17] Hubley et al. concluded that there was no significant difference in image registration accuracy between the different workflows (MRI to CT to CBCT [MRI-CT-CBCT] vs. MRI-CBCT).[18] In an Elekta White Paper, target registration errors were reported to be 0.31–0.54 mm when registering MRI-CBCT and 0.16–0.26 mm for MRI to CT across all intracranial locations for five patients.[19] All these studies were focused on phantom assessment and end-to-end testing with a limited number of clinical patients.

We conducted a retrospective study to evaluate the co-registration accuracy in terms of target coverage for different workflows available on the LGK Icon™ for both frame-based and frameless patients treated in our clinical practice between August 2020 and December 2022. For 50 frame-based (54 targets) and 55 frameless patients (124 targets), we compared the target coverage of the clinically used MRI-CT-CBCT co-registration to the target coverage achieved using the MRI-CBCT co-registration. We designed our study to determine if the MRI-CBCT co-registration workflow is noninferior to the MRI-CT-CBCT workflow.

MATERIALS AND METHODS

In this study, we evaluated two different image registration workflows. In both workflows, the stereotactic space is defined by the CBCT obtained on the LGK Icon™. In the first workflow, the diagnostic MRI was co-registered with a simulation CT, which was, in turn, co-registered with the stereotactic CBCT (MRI-CT-CBCT). In the other workflow, the diagnostic MRI was co-registered with the stereotactic CBCT (MRI-CBCT) [Figure 1]. During the treatment planning process, the radiation oncologist defined all target volumes on the T1 postcontrast MRI. This study did not consider any additional MRI datasets used for treatment planning that were co-registered to the T1 postcontrast MRI, as all volumes were contoured on the T1 postcontrast MRI. All clinical treatment plans were created by the radiation oncologist and approved by a medical physicist prior to treatment delivery. Patients treated previously with the MRI-CT-CBCT workflow were anonymized and re-imported into the LGP treatment planning system to evaluate the dosimetric impact of the different clinical workflows. The anonymized patients were then copied for evaluation with the MRI-CBCT workflow.

Figure 1.

Magnetic resonance imaging (MRI)–computed tomography (CT)–cone-beam computed tomography (CBCT) (a) and MRI-CBCT (b) workflows. Note that all target volumes are defined on the T1 postcontrast MRI, and the CBCT dataset is used as the stereotactic reference dataset for both workflows. MRI: Magnetic resonance imaging, CT: Computed tomography, CBCT: Cone-beam computed tomography

To evaluate the effect on target coverage of removing the CT simulation from the clinical workflow, the target coverage of all clinical targets was recorded for the anonymized patients using the clinically delivered treatment plan. The CT simulation scan was then deleted from the examination of the copied patients, and the T1 postcontrast MRI was co-registered directly with the CBCT stereotactic reference scan for all patients. After co-registration of the T1 postcontrast MRI directly to the CBCT, the target coverage for all targets was recorded for comparison with the clinical plan.

We evaluated 50 frame-based (54 targets) and 55 frameless (124 targets) patients. Patient diagnoses varied based on the clinical patient population at our institution [Table 1]. For the frame-based patients, trigeminal neuralgia and arteriovenous malformation diagnoses were excluded as those cases either did not have targets that were evaluated for coverage or had target volumes drawn on images other than the T1 postcontrast MRI.

Table 1.

Patient diagnoses for cases included in this study

Immobilization	Diagnosis	n
Frame	Vestibular schwannoma	25
	Meningioma	14
	Other schwannoma	4
	Glomus	3
	Pituitary adenoma	3
	Brain metastasis (multiple)	1
	Total	50
Frameless	Brain metastasis (single)	26
	Brain metastasis (multiple)	21
	Meningioma	7
	Glomus	1
	Total	55

Statistical analysis was performed to determine if the MRI-CBCT workflow was noninferior to the MRI-CT-CBCT workflow in terms of target coverage. A 95% confidence interval of the difference in target coverage between the MRI-CBCT and MRI-CT-CBCT was calculated for both the frame-based and frameless workflows, and a noninferior threshold of 2% was chosen such that if the bounds of the 95% confidence interval were within ±2%, the MRI-CBCT workflow was deemed noninferior to the MRI-CT-CBCT workflow. The cases with the largest difference in target coverage were also evaluated to determine the clinical acceptability of the target coverage between the two workflows.

RESULTS

Table 2 provides the difference in target coverage between the MRI-CT-CBCT and MRI-CBCT workflows. The mean target coverage differences for the frame-based and frameless cohorts were −1.20% and −1.06%, respectively.

Table 2.

Statistics for the difference in target coverage for frame-based and frameless patient cohorts

Immobilization	n	Minimum (%)	Maximum (%)	Mean±SD (%)
Frame	54	−5.00	1.00	−1.20±1.43
Frameless	124	−5.00	1.00	−1.06±1.51

SD: Standard deviation

The 95% confidence intervals for the difference in target coverage for the frame-based and frameless patient cohorts are −1.59, −0.81 and −1.32, −0.79, respectively. Since the bounds for 95% confidence intervals are within the predefined noninferiority threshold of ±2%, the MRI-CBCT workflow is deemed noninferior to the MR-CT-CBCT workflow [Table 3].

Table 3.

Noninferiority of the magnetic resonance imaging–cone-beam computed tomography workflow compared to the magnetic resonance imaging–computed tomography–cone-beam computed tomography workflow

Immobilization	Standard workflow	Investigated workflow	Noninferiority threshold (%)	95% CI lower bound (%)	95% CI upper bound (%)	Noninferior
Frame	MRI-CT-CBCT	MRI-CBCT	2.0	−1.59	−0.81	Yes
Frameless	MRI-CT-CBCT	MRI-CBCT	2.0	−1.32	−0.79	Yes

CI: Confidence interval, MRI-CT-CBCT: Magnetic resonance imaging–computed tomography–cone-beam computed tomography, MRI-CBCT: Magnetic resonance imaging–cone-beam computed tomography

Figure 2 shows the frame-based patient with the most significant difference in target coverage between the two workflows. This right vestibular schwannoma target demonstrated target coverage values of 100% on the MRI-CT-CBCT workflow [Figure 2a] compared to 95% on the MRI-CBCT workflow [Figure 2b]. A slight shift in the superior–inferior direction is visible in the sagittal plane [Figure 2b], resulting in decreased target coverage between the workflows. Figure 2 also shows the frameless case with the largest difference in target coverage. A right parietal brain metastasis exhibits coverage values of 98% on the MRI-CT-CBCT workflow [Figure 2c] and 93% on the MRI-CBCT workflow [Figure 2d]. A slight superior–inferior shift is visible in the sagittal plane for this target [Figure 2d].

Figure 2.

Right vestibular schwannoma target with the most significant difference in target coverage between the magnetic resonance imaging (MRI)–computed tomography (CT)–cone-beam computed tomography (CBCT) (a) and MRI-CBCT (b) workflows within the frame-based cohort. Right parietal brain metastasis with the largest difference in target coverage between the MRI-CT-CBCT (c) and MRI-CBCT (d) workflows within the frameless cohort

DISCUSSION

We demonstrated that the MRI-CBCT co-registration workflow is noninferior to the MRI-CT-CBCT workflow for target coverage, suggesting that either approach can be reasonably employed in clinical practice. However, the MRI-CBCT image registration approach offers a faster workflow, eliminating the need for a CT simulation. Reducing the additional radiation from CT scans, though minimal compared to therapeutic doses, helps avoid unnecessary imaging. This reduction adheres to the ALARA principle, enhancing patient safety. Furthermore, for frame-based patients, it reduces the risk of frame movement between CT simulation and treatment.[20] Despite these advantages, there are specific scenarios where a CT scan is helpful. For instance, accurate skull definition cannot be achieved in the presence of MRI artifacts that can compromise definition accuracy.

In addition, inherent distortions in MRI can affect treatment accuracy and significantly affect the coverage of smaller targets in the brain.[21] Clinically, a simulation CT scan can provide supplemental information for target delineation in addition to MRI, particularly for skull base tumors and tumors with known intraosseous extension. To mitigate the risk of target miss, careful planning approaches, including the use of appropriate margins, are essential.

There are some limitations in this study. We did not explore other options for defining the stereotactic reference for frame-based patients, such as using the LGK localizer box. It was not in the scope of this study as we evaluated the clinically utilized MRI-CT-CBCT workflow with the alternative MRI-CBCT workflow. Duggar et al. have shown that the CBCT is reliable in detecting offsets within the stereotactic space and does not sacrifice the accuracy of treatment.[22] Furthermore, the focus was solely on target coverage, without evaluating the impact on organs at risk. The variability in how co-registration was performed, including the region of interest used, also introduces potential inconsistencies. Finally, in our study, treatment planning and dose calculation were performed on the MRI dataset using the TMR-10 algorithm. Since the TMR-10 algorithm does not incorporate heterogeneity corrections, the absence of a CT simulation scan had no significant impact on dose calculation. This further supports the feasibility of the MRI-CBCT workflow as an alternative to MRI-CT-CBCT workflow. It is important to note that a CT simulation scan is necessary for centers using the convolution-based dose algorithm to account for tissue heterogeneity.

CONCLUSION

Based on the study and discussion above, we demonstrated that the MRI-CBCT workflow is noninferior to the MRI-CT-CBCT workflow. With the MRI-CBCT workflow, CT simulation is not required, thus freeing up time on the CT schedule in busy clinics and simplifying the treatment process for patients. Utilizing CBCT to define the stereotactic space allows for a streamlined workflow, leading to efficient treatment sessions and improved patient experience.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.