Phantom-Based Dosimetric Comparison of Helical and Fixed-Beam TomoTherapy for Spatially Fractionated Radiotherapy Using GRID and Lattice Target Designs

Abstract

Introduction

Spatially Fractionated Radiotherapy (SFRT) delivers intentionally heterogeneous dose distributions with alternating high- and low-dose regions and has been widely applied in the management of bulky tumors. In SFRT, high-dose sub-volumes (referred to as vertices) are distributed within the gross tumor volume (GTV) to create a spatial peak–valley dose pattern. However, studies of SFRT using the TomoTherapy platform remain limited. This study evaluates the dosimetric performance of GRID and lattice vertex designs using TomoHelical and TomoDirect techniques.

Methods

A phantom-based planning study was performed with two simulated GTV representing medium- and large-sized tumors. Cylindrical GRID vertices and spherical lattice vertices were created with diameters of 1.00 cm and 1.25 cm and corresponding center-to-center spacings of 3.0 cm and 2.5 cm, respectively. Treatment plans were created using TomoHelical and TomoDirect techniques with identical prescription criteria, requiring at least 50% of the target volume to receive 15 Gy in a single fraction. Beam-on time, normal tissue dose metrics (V30% and V50%), homogeneity index, conformity index, peak-to-edge dose ratio (PEDR), and peak-to-valley dose ratio (PVDR) were evaluated. Delivery accuracy was assessed using ArcCHECK measurements with a 3%/2 mm gamma criterion.

Results

All plans met prescription and delivery accuracy requirements, with gamma passing rates exceeding 95%. TomoHelical produced higher PVDR and PEDR values, improved conformity, and lower V50% compared with TomoDirect. TomoDirect achieved shorter beam-on times but showed greater variability in vertex mean dose. Lattice configurations yielded higher PVDR values than GRID, while vertex diameter had minimal impact on most dosimetric parameters.

Conclusion

TomoHelical delivery combined with lattice designs provided superior dose modulation and normal tissue sparing for SFRT, while requiring longer delivery times. In contrast, GRID designs enabled faster treatment delivery. These findings provide practical guidance for optimizing SFRT planning using the TomoTherapy platform.

Introduction

Advancements in radiotherapy techniques aimed at enhancing tumor control while reducing treatment-related toxicity, significant challenges remain in the management of bulky tumors, lesions in close proximity to critical organs, and radioresistant disease. In such cases, the delivery of sufficiently high radiation doses is often limited by normal tissue tolerance, potentially compromising treatment effectiveness.

Spatially Fractionated Radiotherapy (SFRT) has emerged as an alternative approach to address these limitations. Unlike conventional uniform dose delivery, SFRT intentionally generates highly non-uniform dose distributions within the tumor, creating alternating regions of high dose (peaks) and low dose (valleys). ¹ This strategy enables selective dose escalation to intratumoral subvolumes while reducing the integral dose to surrounding normal tissues. Clinical and dosimetric studies have demonstrated that SFRT can achieve effective high-dose delivery with acceptable toxicity profiles in both palliative and selected curative settings.^2-4 However, in clinical practice, SFRT is frequently delivered in combination with conventional radiotherapy. ⁵

The therapeutic potential of SFRT is supported by proposed biological mechanisms, including radiation-induced bystander effects, vascular damage, and immune modulation, which may enhance tumor response even in radioresistant lesions.^6-9 While these biological effects are increasingly recognized, the successful clinical implementation of SFRT remains strongly dependent on treatment delivery techniques and their ability to generate and maintain robust peak–valley dose patterns.^10-12

The application of SFRT involves specific target designs that define the geometric arrangement of high-dose sub-volumes, referred to as vertices, within the gross tumor volume (GTV), rather than the anatomical delineation of the tumor itself. These vertices represent localized regions intentionally prescribed with higher radiation doses to generate the characteristic peak–valley dose pattern of SFRT. In this study, two commonly used SFRT target designs were investigated: GRID radiotherapy and lattice radiotherapy (LRT). GRID radiotherapy delivers radiation through cylindrical high-dose channels and has been adapted for use with Helical Tomotherapy via multileaf collimator (MLC) modulation to create patient-specific spatially modulated dose patterns within the tumor. ¹³ Alternatively, LRT extends the SFRT concept into three dimensions by distributing spherical high-dose vertices throughout the tumor volume, making it particularly suitable for very large lesions.^14,15 When implemented using Helical Tomotherapy, LRT has been shown to achieve high peak to valley dose ratios, although treatment delivery time tends to increase with larger vertex size and spacing. ¹⁶

Although both GRID and LRT show promise for treating bulky tumors, systematic evaluation of their dosimetric performance across different TomoTherapy delivery techniques remains limited. Helical Tomotherapy (TomoHelical) delivers radiation through continuous gantry rotation with a fan-beam geometry, resulting in highly uniform dose distributions along the superior-inferior direction. In contrast, TomoDirect is a fixed beam delivery mode that utilizes multiple static gantry angles without continuous rotation. Owing to its reduced delivery complexity, TomoDirect may reduce beam-on time; however, differences in beam orientation and delivery mechanics may lead to distinct spatial dose characteristics and peak-valley modulation compared with TomoHelical.

Currently, no comprehensive study has directly compared the dosimetric characteristics of SFRT delivered using TomoHelical and TomoDirect with both cylindrical GRID and spherical lattice vertex designs. Characterizing how delivery technique and vertex geometry influence peak-valley modulation, normal tissue dose, and treatment efficiency is essential for optimizing the clinically feasible implementation of SFRT on TomoTherapy platforms.

Therefore, the purpose of this study is to evaluate the dosimetric implications of GRID and lattice target designs using TomoHelical and TomoDirect delivery techniques for SFRT. Clinically relevant dosimetric parameters related to spatial dose modulation, normal tissue sparing, treatment efficiency, and delivery accuracy are evaluated to provide practical guidance for implementing SFRT in clinical TomoTherapy workflows.

Methods and Materials

Phantom and Target Design

Computed tomography (CT) images of a Cheese Phantom (Tomo® phantom HE, Sun Nuclear Corporation, Melbourne, FL, USA) were acquired using a CT simulation protocol with a slice thickness of 1 mm. The resulting DICOM datasets were imported into MIM Maestro® (MIM Software Inc., Cleveland, OH, USA) for structure simulation.

Two cylindrical GTVs were simulated on phantom images to represent different tumor sizes. The medium-sized tumor (GTV1) was contoured as a cylinder with an 8 cm diameter and a 6 cm length (Figure 1A), while the large-sized tumor (GTV2) was contoured with a diameter of 12 cm and a length of 9 cm (Figure 1B). To assess dose fall-off near the tumor boundary, an Edge structure was generated by expanding 2 mm outward from the GTV surface. In addition, a Normal Tissue structure was defined as a 5-cm isotropic expansion from the GTV, excluding the GTV itself, to evaluate dose deposition in the surrounding healthy tissue (Figure 1).

Figure 1.

Definition of simulated tumor and evaluation structures used in Cheese phantom (diameter: 30 cm, length: 18 cm). (A) Medium-sized cylindrical tumor volume (GTV1) with the corresponding Edge and Normal Tissue structures. (B) Large-sized cylindrical tumor volume (GTV2) with the corresponding Edge and Normal Tissue structures

Two SFRT target designs were manually created and arranged in a square pattern to achieve a relatively uniform spatial distribution within each GTV, as illustrated in Figure 2.

Figure 2.

Illustration of SFRT target designs. (A) Cylindrical GRID targets within the medium-sized tumor volume (GTV1). (B) Spherical lattice targets within GTV1. (C) Cylindrical GRID targets within the large-sized tumor volume (GTV2). (D) Spherical lattice targets within GTV2

Target Design A: Cylindrical GRID Vertices (GRID)

Cylindrical vertices were designed to mimic the geometry of a virtual physical GRID block, using MLC modulation to shape the radiation beam accordingly (Figure 2A and C). Two parameter sets were evaluated.

- Cylinders with a 1.0-cm diameter and a 3.0-cm center-to-center spacing

- Cylinders with a 1.25-cm diameter and a 2.5-cm center-to-center spacing

Target Design B: Spherical Lattice Vertices (LRT)

Spherical vertices were contoured as three-dimensional vertices distributed throughout the GTV (Figure 2B and D). Two parameter sets were evaluated.

- Spheres with a 1.0-cm diameter and a 3.0-cm center-to-center spacing

- Spheres with a 1.25-cm diameter and a 2.5-cm center-to-center spacing

All vertex structures were combined into a single structure, referred to as Target_all, for treatment planning and dosimetric evaluation. To enable consistent geometric comparison, the number of high-dose vertices was kept constant within each target design for a given GTV (i.e.,cylinders for GRID and spheres for lattice). Due to differences in geometric shape and spatial arrangement between cylindrical and spherical vertices, the total number of vertices was not necessarily identical between GRID and lattice designs. Instead, the configurations were designed to achieve appropriate high-dose coverage within the GTV. Although variations in vertex size and spacing resulted in minor differences in the distance between some vertices and the GTV boundary, all vertices were fully contained within the GTV. For each high-dose vertex, two concentric ring structures were generated for valley dose control. Ring 1 was created with a 5 mm expansion from the vertex edge, and Ring 2 with an additional 5 mm expansion from Ring 1.

Treatment Planning

All target structures were imported into the Precision® Treatment Planning System (Accuray Inc., Sunnyvale, CA, USA) for treatment planning using both TomoHelical (TH) and TomoDirect (TD) delivery techniques. Planning parameters were selected to ensure clinically feasible delivery while maintaining appropriate spatial dose modulation for SFRT. The final dose was calculated using the convolution/superposition algorithm with a grid size of 0.99 × 1.00 × 0.99 mm³. The optimization consisted of three iteration rounds (60 iterations in total). The modulation factor was constrained to a maximum value of 2.0, and the actual modulation factors ranged from 1.1 to 1.7.

For TomoHelical delivery, a field width of 5 cm with Dynamic Jaw mode was used. The pitch was set to 0.05 for Target Design A, cylindrical GRID vertices (GRID_TH), and 0.1 for Target Design B, spherical lattice vertices (LRT_TH), reflecting differences in target geometry and modulation requirements. Pitch values were selected to maintain a gantry period of less than 60 s, ensuring plan deliverability without compromising dose modulation quality. This parameter selection was intended to closely reflect routine clinical workflow in TomoTherapy-based treatments.

For TomoDirect delivery, a field width of 5 cm with Dynamic Jaw mode was used for both Target Design A (cylindrical GRID vertices, GRID_TD) and Target Design B (spherical lattice vertices, LRT_TD). To achieve balanced dose coverage around the centrally located GTV, coplanar beam angles were selected using approximately uniform angular spacing over 360°, resulting in near-equally distributed beam directions (≈360°/N, where N is the number of beams). Specifically, seven beams (∼51° spacing) were used for GTV1 and nine beams (∼40° spacing) for GTV2. This approach was adopted to provide quasi-rotational dose coverage for the centrally located target within the symmetric phantom, while maintaining a balance between adequate angular sampling and avoiding excessive beam overlap. The initial gantry angle (0°) and subsequent evenly spaced increments were chosen to ensure reproducibility and to avoid directional bias.

The optimization objectives for the Target_all were defined to ensure adequate dose coverage of the Target_all structure. Specifically, a minimum dose of 15 Gy (prescription dose) was required for at least 50% of the Target_all volume, while a minimum dose of 14.25 Gy (95% of the prescription dose) was required for 100% of the volume. Additional optimization constraints were applied to surrounding normal tissue structures and a ring structure to control dose spillage and maintain the intended peak–valley dose pattern. Optimization objectives were adjusted iteratively during planning to achieve the desired dosimetric characteristics.

Dosimetric Evaluation

Paired comparisons were performed under identical planning conditions using both dosimetric and treatment delivery metrics. Evaluated parameters included beam-on time (BOT), V30% and V50% were defined as the percentage volume of the normal tissue structure receiving ≥30% and ≥50% of 15 Gy, the mean dose (Dmean) of individual high-dose vertices within Target_all, and the homogeneity index (HI), as given in Equation (1), ¹⁷ and the conformity index (CI), as given in Equation (2), ¹⁸ which were calculated based on the Target_all structure using the prescription isodose corresponding to the D50% criterion (15 Gy).

$\text{HI}=\frac{{\mathrm{D}}_{2\%}-{\mathrm{D}}_{98\%}}{{\mathrm{D}}_{50\%}}$

(1)where D_2%, D_98% and D_50% represent the doses received by 2%, 98%, and 50% of the Target_all volume, respectively.

$\text{CI}=\frac{{{\text{TV}}_{\text{PIV}}}^2}{\text{TV}\hspace{0.17em}\mathrm{x}\hspace{0.17em}\text{PIV}}$

(2)where TV is the Target_all volume, PIV is the volume of external contour which received 14.25 Gy, and TV_PIV is the target volume covered by 14.25 Gy.

For SFRT-specific assessment, additional dose modulation indices were calculated, including the peak-to-edge dose ratio (PEDR) and the peak-to-valley dose ratio (PVDR). PEDR was defined as the ratio of the mean dose delivered to the Target_all structure to the mean dose within the edge structure (defined as a 2-mm isotropic expansion from the GTV surface), as given in Equation (3). PVDR was defined as the ratio of the mean dose delivered to the Target_all structure to the mean dose within the valley region of the GTV, defined as the GTV excluding Target_all, as defined in Equation (4). A higher PVDR indicates greater contrast between peak and valley dose regions, reflecting enhanced spatial dose modulation and improved spatial dose quality characteristic of SFRT.

$\text{PEDR}=\frac{{\mathrm{D}}_{\text{mean}}\left(\text{Target}\_\text{all}\right)}{{\mathrm{D}}_{\text{mean}}\left(\text{Edge}\right)}$

(3)

$\text{PVDR}=\frac{{\mathrm{D}}_{\text{mean}}\left(\text{Target}\_\text{all}\right)}{{\mathrm{D}}_{\text{mean}}\left(\text{Valley}\right)}$

(4)where Dmean(Target_all) is the mean dose of Target_all, Dmean(Edge) is the mean dose within the Edge structure, and Dmean(Valley) is the mean dose within the GTV excluding Target_all.

Dose Verification

Plan-specific quality assurance measurements were performed using the ArcCHECK phantom (Sun Nuclear Corporation, Melbourne, FL, USA) and delivered on a Radixact X9 system (Accuray Inc., Sunnyvale, CA, USA). Plan delivery accuracy was evaluated using gamma (γ) index analysis with a 3%/2 mm acceptance criterion. Gamma analysis was performed using global normalization with a 10% dose threshold. The ArcCHECK detector spacing was 1 cm. The global maximum dose was used as the reference for normalization. All evaluated plans achieved gamma passing rates of at least 95%, in accordance with the recommendations of AAPM Task Group 218, ¹⁹ confirming the accuracy and deliverability of the planned SFRT dose distributions.

Statistical Analysis

All quantitative dosimetric data were assessed for normality using the Shapiro–Wilk test. As the evaluated variables did not satisfy the assumption of normal distribution, paired comparisons were performed using the nonparametric Wilcoxon signed-rank test. A two-sided p-value < 0.05 was considered statistically significant. All statistical analyses were conducted using IBM SPSS Statistics (version 27; IBM Corp., Armonk, NY, USA). This study was conducted using a phantom model and did not involve human participants, patient data, or biological materials. Therefore, ethical approval and patient consent were not required.

Results

Dosimetric and Delivery Characteristics

Table 1 summarizes the dosimetric and delivery parameters for all treatment plans. The gantry period for TomoHelical plans ranged from 40.7 to 58.3 s and remained below the 60 s limit, indicating that all plans were clinically feasible for treatment delivery.

Table 1.

Dosimetric and Delivery Parameters of Spatially Fractionated Radiotherapy Plans Using TomoHelical and TomoDirect Techniques

Delivery technique	GTV	Target design	Target diameter (cm)	Gantry period (s)	D50% (Gy)	PVDR	Beam-ontime (s)	Gamma passing rate (%)
TomoHelical	GTV1	Lattice	1.25	58.3	15.29	2.88	1124.8	98.9
			1.00	46.0	15.38	3.28	1044.9	100
		GRID	1.25	48.6	15.65	2.46	842.4	100
			1.00	48.5	15.69	2.53	851.1	100
	GTV2	Lattice	1.25	52.9	15.09	2.13	1589.1	97.6
			1.00	44.5	15.18	2.30	1569.7	99.3
		GRID	1.25	47.0	15.17	1.75	1107.7	99.4
			1.00	40.7	15.13	1.75	983.2	98.4
TomoDirect	GTV1	Lattice	1.25	—	15.01	2.62	713.8	100
			1.00	—	15.06	2.90	825.7	99.6
		GRID	1.25	—	15.08	2.34	724.7	100
			1.00	—	15.11	2.43	641.9	100
	GTV2	Lattice	1.25	—	15.03	2.02	1237.8	97.8
			1.00	—	15.03	2.23	1410.3	98.0
		GRID	1.25	—	15.00	1.71	849.7	97.7
			1.00	—	15.01	1.77	845.9	98.1

Abbreviations: GRID = Target Design A (cylindrical GRID targets); Lattice = Target Design B (spherical lattice targets). GTV1 and GTV2 represent medium- and large-sized simulated tumor volumes, respectively. Gamma passing rates were evaluated using an absolute dose criterion of 3%/2 mm.

All planning configurations achieved the prescription requirement, with D50% values exceeding 15 Gy across all plans, indicating that at least 50% of the target volume (Target_all) received the prescribed dose, consistent with the intended SFRT dose escalation strategy.

Beam-on time demonstrated clear differences between delivery techniques, target designs, and tumor sizes. For GTV1, the shortest beam-on time was observed for the 1.00-cm GRID_TD plan (641.9 s), whereas the longest beam-on time occurred for the 1.25-cm LRT_TH plan (1124.8 s). Similarly, for GTV2, the shortest beam-on time was achieved with the 1.00-cm GRID_TD plan (845.9 s), while the longest was observed for the 1.25-cm LRT_TH plan (1589.1 s). Overall, TomoDirect consistently resulted in shorter beam-on times than TomoHelical for both tumor sizes. In contrast, TomoHelical delivery, lattice design (Target Design B), and larger vertex diameter (1.25 cm) were associated with increased beam-on times.

Peak–Valley Dose Modulation

PVDR showed a clear dependence on tumor size. For GTV1, PVDR values ranged from 2.34 to 3.28, whereas for GTV2, PVDR values decreased to a range of 1.71 to 2.30. This trend indicates that PVDR decreased with increasing tumor volume across delivery techniques and target designs. Such behavior is consistent with a characteristic feature of SFRT, in which achieving pronounced peak–valley dose modulation becomes increasingly challenging as tumor size increases.

Delivery Accuracy

Delivery quality assurance results demonstrated that all treatment plans were accurately delivered. Gamma (γ) index analysis using the 3%/2 mm criterion yielded passing rates greater than 95% for all plans (Table 1). These results indicate good agreement between calculated and measured dose distributions and support the feasibility of delivering SFRT dose patterns using the TomoTherapy platform.

Paired Comparison of Dosimetric Parameters

Figure 3 illustrates paired dot plots comparing dosimetric and delivery related parameters generated under identical planning conditions, stratified by delivery technique (TomoHelical vs. TomoDirect), target design (Lattice vs. GRID), and vertex diameter (1.25 cm vs. 1.00 cm).

Figure 3.

Paired dot plots comparing delivery techniques (TomoHelical vs TomoDirect), target designs (Lattice vs GRID), and vertex diameters (1.25 cm vs 1.00 cm) for (A) beam-on time, (B) V30% of normal tissue structure, and (C) V50% of normal tissue structure. Thin gray lines connect paired plans under identical conditions, and black diamonds represent group medians. Statistical significance was assessed using the Wilcoxon signed-rank test

Beam-On Time

As shown in Figure 3A, TomoDirect demonstrated significantly shorter beam-on times than TomoHelical (p = 0.008). In addition, GRID vertices required significantly less delivery time than lattice vertices (p = 0.016), whereas no significant difference was observed between vertex diameters (p = 0.844).

Normal Tissue Dose

At low-dose levels, no statistically significant differences were observed in V30% among delivery techniques, target designs, or vertex diameters (Figure 3B). In contrast, high-dose exposure (V50%) was significantly lower for TomoHelical compared with TomoDirect (p = 0.012) and for GRID compared with lattice designs (p = 0.028), indicating improved normal tissue sparing at higher dose levels (Figure 3C). Vertex diameter did not significantly affect V50%.

Dose Homogeneity and Conformity

TomoHelical delivery and lattice designs yielded significantly lower HI values, indicating more homogeneous dose distributions (p = 0.034 and p = 0.018, respectively; Figure 4A). No significant effect of target diameter on HI was observed. For conformity, TomoHelical demonstrated significantly higher CI values compared with TomoDirect (p = 0.012; Figure 4B), whereas target design and target diameter did not significantly influence CI.

Figure 4.

Paired dot plots comparing dose homogeneity, conformity, and SFRT-specific dose modulation indices across delivery techniques, target designs, and target diameters. Comparisons are shown for (A) homogeneity index (HI), (B) conformity index (CI), (C) peak-to-edge dose ratio (PEDR), and (D) peak-to-valley dose ratio (PVDR). Statistical significance was assessed using the Wilcoxon signed-rank test

SFRT Specific Dose Modulation

PEDR was significantly higher for TomoHelical compared with TomoDirect (p = 0.035) and for lattice compared with GRID designs (p = 0.012), with no significant dependence on vertex diameter (Figure 4C). Similarly, PVDR was significantly higher for TomoHelical delivery (p = 0.017), lattice designs (p = 0.012), and the 1.00-cm diameter configuration (p = 0.018), reflecting enhanced peak–valley dose modulation (Figure 4D).

Target Dose Uniformity

As shown in Figure 5, box plots illustrate the Dmean delivered to individual target vertices across all treatment plans. All plans were generated using the same prescription criterion (D50% = 15 Gy to Target_all), ensuring a consistent normalization baseline. TomoDirect exhibited a broader distribution and greater variability in vertex Dmean values across both tumor sizes and target designs, indicating increased heterogeneity in the dose delivered to individual vertices with the fixed-beam TomoDirect technique. In contrast, TomoHelical demonstrated more consistent dose distributions across vertices, reflecting improved uniformity in dose delivery to SFRT vertices.

Figure 5.

Box plots of the mean dose (Dmean) delivered to Target_all for all treatment plans

Directional Dose Characteristics

Figure 6 illustrates representative dose distributions in both axial and sagittal views, highlighting the effects of vertex diameter, center-to-center spacing, and target design on spatial dose modulation. In the lattice vertices with a larger center-to-center spacing of 3.0 cm (1.00-cm diameter) produced more pronounced low-dose valleys compared with those using a reduced spacing of 2.5 cm (1.25-cm diameter) as show in Figure 6A and C. For GRID vertices, the spacing variation primarily occurs along the lateral direction, resulting in more pronounced low-dose regions in this direction, as show in Figure 6B and D. This characteristic reflects the increased separation between high-dose vertices, allowing for greater valley dose reduction.

Figure 6.

Representative dose distributions for lattice and GRID target designs shown in both axial (left) and sagittal (right) views. (A and B) Vertex diameter of 1.00 cm with 3.0-cm center-to-center spacing. (C and D) Vertex diameter of 1.25 cm with 2.5-cm center-to-center spacing. Panels (A and C) correspond to lattice designs, while (B and D) correspond to GRID designs. Isodose levels are displayed using a consistent color scale across all plans

The sagittal views further demonstrate that valley dose reduction was less pronounced in directions perpendicular to the superior–inferior (SI) axis compared with along the SI direction. This directional dependence is primarily attributable to the delivery characteristics of TomoTherapy, in which dose modulation is dominated by continuous gantry rotation combined with couch translation along the SI axis.

For GRID designs (Figure 6B and D), dose distributions exhibited greater extension along the SI direction compared with lattice designs. As a result, increased dose spillage outside the GTV was observed, particularly along the beam delivery direction.

Discussion

This study evaluated the feasibility and dosimetric performance of SFRT on the TomoTherapy platform. While previous investigations have demonstrated the potential of TomoTherapy for SFRT delivery,^13,16,20 the present work extends existing knowledge by directly comparing two delivery techniques TomoHelical and TomoDirect together with two commonly employed SFRT target designs, lattice and GRID. The results provide additional insight into the dosimetric characteristics and clinical practicality of SFRT delivery using TomoTherapy systems.

The findings confirmed that both TomoHelical and TomoDirect are capable of generating effective peak–valley dose patterns. However, TomoHelical consistently produced higher PVDR values, indicating superior peak–valley dose contrast. In addition, TomoHelical demonstrated higher PEDR values and lower V50% of normal tissue, suggesting improved sparing of surrounding healthy tissues. More favorable homogeneity and conformity indices were also observed with TomoHelical compared with TomoDirect. It should be noted that HI in this study reflects dose uniformity within individual vertices rather than overall tumor dose heterogeneity, which is more appropriately characterized by PVDR in the context of SFRT. These dosimetric advantages can be attributed to the continuous rotational delivery of TomoHelical, which facilitates smoother longitudinal dose modulation and reduces high-dose spillage along the superior–inferior direction through dynamic jaw motion.

In contrast, TomoDirect achieved shorter beam-on times, which may offer advantages in time-constrained clinical settings. However, its overall dosimetric performance was inferior to that of TomoHelical. It should be noted that the reported beam-on times for TomoDirect do not include additional time associated with inter-beam gantry rotation and couch translation. Consequently, actual clinical delivery time may be underestimated for TomoDirect, and the practical time-saving benefit may be less pronounced during treatment delivery than suggested by planning system estimates. Therefore, beam-on time should be interpreted as radiation-on time rather than total in-room treatment time, particularly for TomoDirect.

Comparison between lattice and GRID designs revealed that lattice designs achieved significantly higher PVDR values, reflecting a more pronounced peak–valley dose pattern, which is a defining feature of SFRT. This observation is consistent with previous studies. ²¹ The three-dimensional geometric arrangement of lattice vertices allows for wider valley regions between high-dose vertices. When combined with fan-beam delivery and dynamic jaw modulation, as in TomoHelical, unwanted dose deposition particularly along the superior–inferior direction is further reduced.^16,22 As a result, lattice designs also demonstrated higher PVDR, PEDR values and lower V50%, indicating reduced high-dose spillage into normal tissue. GRID designs, however, were associated with shorter beam-on times and simpler contouring requirements, as GRID structures are easier and less time-consuming for medical physicists to delineate, potentially improving efficiency in routine clinical workflows.

A recent study by Setianegara et al. ²³ comparing proton-based pGRID and pLATTICE techniques reported substantially higher PVDR values for pGRID, attributable to target alignment along the proton beam path and the sharp dose gradients achievable with proton collimation. In this study, PVDR values obtained with photon-based TomoTherapy were lower, reflecting the inherent physical limitations of photon beams compared with charged particles. Nevertheless, the PVDR range observed here (1.71–3.28) is consistent with previously reported photon-based SFRT studies, including the TomoTherapy-GRID work by Sheikh et al. ²⁰ which employed similar vertex diameters and separations. In that study, VMAT-based GRID delivery produced lower PVDR values, whereas TomoTherapy-based GRID achieved PVDR levels comparable to those observed in the present study. The PEDR values in this study (3.13–7.33) were relatively higher, potentially reflecting differences in vertex-to-edge separation and vertex placement.

To evaluate the impact of vertex diameters (1.00 and 1.25 cm), the design configurations were designed with consideration of both clinical feasibility and the quality of spatial dose modulation inherent to SFRT. Vertices with a diameter of 1.00 cm and a center-to-center spacing of 3.0 cm were selected to achieve clinically feasible beam-on times while maintaining an effective peak–valley dose contrast. This geometric configuration is consistent with previous reports,^16,20 in which valley-to-peak dose ratio (VPDR) values below 0.4 were achieved, indicating pronounced spatial dose modulation and a favorable therapeutic ratio in SFRT, particularly when implemented using TomoTherapy systems. In addition, the 1.25-cm diameter vertices were assigned a center-to-center spacing of 2.5 cm, with the intent of matching both the high-dose region size and the end-to-end distance between adjacent vertices to 1.25 cm. This configuration corresponds to the effective geometric opening of two MLC leaves (0.625 cm per leaf). This design approach was intended to produce high-dose and very-low-dose regions with dose profiles comparable to those generated by MLC-based intensity modulation, thereby potentially reducing beam-on time and enhancing clinical deliverability.

The results demonstrated no statistically significant differences in beam-on time, V30%, V50%, HI, CI, and PEDR between the two vertex diameters. However, limitations in CT simulation resolution and treatment planning voxel size may introduce partial-volume and discretization effects, leading to minor deviations between the nominal and realized vertex diameters. Although higher PVDR values were observed for the 1.00-cm diameter vertices, this effect is more likely attributable to the larger center-to-center spacing, which permits wider valley regions, rather than to vertex diameter alone. Because vertex diameter was evaluated in conjunction with different center-to-center spacings, the observed PVDR differences likely reflect spacing-driven changes in valley geometry. Furthermore, due to the helical delivery geometry, MLC opening width does not directly translate to the physical dimensions of the vertex structures. Consequently, given the relatively small difference between the two evaluated diameters, substantial dosimetric differences were not anticipated.

A strength of this study is the simultaneous evaluation of delivery techniques and SFRT target design under controlled geometric conditions. In addition, delivery quality assurance measurements demonstrated gamma passing rates exceeding 95% for all plans, confirming the ability of the TomoTherapy system to accurately deliver spatially heterogeneous SFRT dose distributions.

Several limitations should be acknowledged. First, the phantom-based design does not represent the anatomical complexity encountered in clinical practice, such as irregular tumor geometries or proximity to organs at risk. Second, different pitch values were used between target designs to maintain clinically feasible delivery conditions, which may introduce a residual confounding effect on dose modulation metrics. In addition, TomoHelical delivery is limited to coplanar beam geometry and does not allow non-coplanar beam arrangements, which may restrict achievable degrees of freedom for spatial dose modulation compared with some linac-based techniques.

Despite these limitations, the controlled phantom geometry enabled systematic evaluation of key technical factors influencing SFRT dose modulation and treatment delivery characteristics. These findings therefore provide technical guidance for SFRT treatment planning on the TomoTherapy platform.

In clinical practice, additional constraints such as irregular tumor geometry, proximity to organs at risk, and patient-specific anatomical variations may influence achievable dose distributions. Furthermore, the relatively long delivery times observed for some plans may increase sensitivity to intrafraction motion. Although patient immobilization and image guidance can help mitigate motion during treatment, the impact of motion in SFRT is expected to manifest as dose blurring between peak and valley regions, potentially reducing the PVDR. Previous work has shown that lattice dose distributions remain relatively stable up to approximately 0.5 cm of motion for 1 cm sphere configurations, while larger motion amplitudes may lead to degradation of PVDR. ²⁴ Given that the center-to-center spacing in this study (2.5–3.0 cm) is larger than typical residual motion under clinical conditions, the overall spatial dose modulation is expected to be preserved, although some reduction in peak–valley contrast may occur. Nevertheless, further investigation is required to quantitatively assess the robustness of these dose distributions under patient-specific motion conditions.

Finally, this study focused on dosimetric characteristics and did not evaluate biological responses or tumor–normal tissue interactions associated with SFRT dose modulation. Future investigations incorporating patient datasets, biological modeling, and clinical outcome analyses will be necessary to further define the clinical role of TomoTherapy-basedSFRT.

Conclusion

This phantom-based study demonstrates that TomoTherapy can generate spatially fractionated dose distributions using both TomoHelical and TomoDirect delivery techniques. TomoHelical combined with lattice designs produced higher peak–valley dose modulation and reduced high-dose exposure to surrounding tissue, whereas GRID designs enabled shorter delivery times. These findings provide technical insight for SFRT treatment planning using TomoTherapy. Further investigations incorporating patient datasets are required to evaluate clinical feasibility and outcomes.