Abstract
Background/Aim: Patients with meningeosis carcinomatosa or primary brain tumors may require craniospinal irradiation. In craniospinal volumetric modulated arc therapy (VMAT-CSI), field junctions are sensitive regions prone to underdosage. This study investigated whether global, planning target volume (PTV)-based quality metrics can detect such localized deficits and how different field-overlap widths affect junction dosimetry.
Patients and Methods: Retrospective computed tomography datasets from six patients were analyzed. For each patient, VMAT-CSI plans with overlap lengths of 4, 6, 8, 10, and 12 cm were generated. Dose coverage (D98%) and homogeneity index (HI) were evaluated for the PTV and a junction-focused subvolume (PTV-VOI). Conformity indices (RTOG, Salt–Lomax, Lomax, and Van’t Riet) were calculated for both structures.
Results: Increasing overlap improved D98% for both PTV and PTV-VOI. While the PTV often met D98% ≥95% at small overlaps, adequate PTV-VOI coverage was achieved only at ≥10 cm. VOI-based RTOG and Salt–Lomax indices increased with overlap, the Lomax index decreased slightly, and Van’t Riet remained stable. Global PTV-based indices showed minimal variation.
Conclusion: Adequate dose coverage at the field junctions in VMAT-CSI treatment is achieved with field overlaps of 10-12 cm. Since global PTV metrics mask local underdosages, clinicians should use VOI assessments at the field junctions rather than relying solely on global plan indices.
- Meningeosis carcinomatosa
- primary brain tumor
- craniospinal irradiation
- volumetric modulated arc therapy
- optimization
Introduction
Craniospinal irradiation (CSI) remains a cornerstone in the treatment of central nervous system malignancies such as medulloblastoma and other embryonal tumors (1). The technical challenge of CSI lies in delivering a homogeneous dose across the entire craniospinal axis, which spans a large and anatomically complex region. In particular, achieving seamless dose coverage at the junctions between cranial and spinal fields is critical, as improper alignment or modulation can result in cold spots (underdosage) or hot spots (overdosage), thereby compromising treatment efficacy and increasing toxicity risks (2-4).
Volumetric modulated arc therapy (VMAT) has emerged as a preferred technique for CSI due to its superior dose conformity, improved treatment efficiency, and reduced exposure of organs at risk compared to conventional 3D conformal or intensity-modulated radiotherapy (IMRT) approaches. Nevertheless, the intrinsic challenge of managing dose gradients at field junctions persists. The width of the overlap region between adjacent fields plays a decisive role in achieving dosimetric robustness. Narrow overlaps tend to produce steep dose gradients and increase sensitivity to positional inaccuracies, while broader overlaps enable smoother dose transitions and mitigate the effects of setup uncertainty (5-14).
Recent investigations have highlighted the relevance of optimizing field overlap regions in VMAT-based CSI. In 2023, Zhang et al. proposed a feasibility optimization strategy in a multi-isocenter VMAT-CSI framework, demonstrating improved dose homogeneity and conformity through targeted management of field junctions (15). Maddalo et al. showed that the use of an automatic feathering (AF) algorithm to the overlap technique significantly improves plan robustness, particularly for larger overlap lengths, and that AF-optimized plans achieve robustness comparable to plans generated using more sophisticated and time-consuming approaches such as gradient optimization (16). These studies underscore the clinical importance of overlap region design in VMAT-CSI but also reveal a lack of standardized guidance regarding the optimal extent of such overlaps.
Even though substantial progress has been made in the field of craniospinal irradiation, scientific data on a systematic investigation of the influence of field overlap length on dose quality and the resulting dosimetric parameters remains limited. Dosimetric parameters generally refer to the entire volume to be irradiated. However, the underdoses in the field transitions are spatially limited and small in relation to the total volume. Since, as mentioned above, these impair clinical efficacy, the question arises as to whether global quality indicators – i.e., parameters that refer to the entire PTV or large volumes – are sufficiently sensitive to reliably detect these locally limited dose deficits.
Our study therefore addresses this issue on two levels: On the one hand, we examine the extent to which the length of field overlap influences the occurrence and extent of such cold spots in the field transition regions. On the other hand, we apply quality indicators for assessing dose conformity both globally and locally and compare them with each other. We expect to be able to determine whether there is an optimal size range for field overlap width and how this can be measured or even controlled using quality parameters. Ultimately, the findings aim to clarify whether established global quality metrics remain adequate for VMAT-CSI or whether locally responsive evaluation methods are required to maintain high clinical standards.
Patients and Methods
Six patients who had been treated in our Department were retrospectively selected for this study. Each patient underwent non-contrast-enhanced planning computed tomography (CT) scans with a slice thickness of 3 mm. Imaging was performed using a 40-slice CT scanner (Biograph mCT, Siemens AG, Erlangen, Germany). Planning CT images were transferred to the ARIA® oncology information system (Version 16.0, Varian Medical Systems, Palo Alto, CA, USA) for imaging processing and target volume delineation. Dose calculations were performed using the Eclipse™ treatment planning system (TPS) (Version 16.1, Varian Medical Systems).
Treatment planning. VMAT plans were generated for each patient with field overlap widths of 4, 6, 8, 10, and 12 cm. All plans were designed using three isocenters (IC1-IC3). The IC coordinates were calculated with the Arc Setup Tool in the TPS and adjusted to ensure the x-coordinate was set to zero. The z-coordinate was rounded to an integer value, maintaining a distance of 20 cm between adjacent ICs. The y-coordinates were aligned to consistent integer values for all patients. A single dose of 1.8 Gy was applied to the PTV in 20 fractions, resulting in a total dose of 36 Gy. The treatment plans were accepted when reaching a PTV Dmean=100% while Dmax≤107%.
Plan comparison parameters. The quality of dose distribution was assessed using the following criteria: (i) D98% ≥ 95% (dose delivered to 98% of the target volume), (ii) Conformity Indices (CI) according to Salt Lomax, Lomax, RTOG, and Van’t Riet, (iii) Homogeneity Index (HI). The calculation methods for the respective CI and HI values are summarized in Figure 1.
To enable a targeted analysis of the field junction region, the PTV-based metrics listed in Figure 1 were adapted in a second step to a spatially restricted PTV subvolume (Figure 2). For this purpose, a volume of interest (VOI) was defined as a reference structure encompassing the field overlap region as shown in Figure 3. The VOI served exclusively to spatially delineate the region in which junction-related dose deficits occurred and was not used as a dosimetric evaluation volume. The VOI had a fixed cranio–caudal extent of 34 CT slices for all patients, corresponding to a longitudinal length of 10.2 cm. Its position was individually adapted to the location of underdosage regions within the cranial field junction (IC1) (Figure 4A-E).
Volume of interest (VOI) definition in radiotherapy. Within the body outline, the planning target volume (PTV, red) is shown, from which a subvolume is defined as PTV-VOI (yellow). Based on its longitudinal extent, an additional VOI with identical length is defined within the body.
Enlarged dose distributions in the field junction region for patient #1 at increasing field overlap lengths: (A) 4 cm, (B) 6 cm, (C) 8 cm, (D) 10 cm, and (E) 12 cm. Dose is shown as a color-wash representation ranging from 95% of the prescribed dose to Dmax. The junction-focused subvolume of the planning target volume (PTV-VOI) is shown in yellow. With increasing overlap length, dose gaps in the junction region decrease and the dose distribution becomes more homogeneous.
The portion of the PTV located within this VOI was subsequently defined as the PTV-VOI and served as the target structure for the VOI-based dosimetric analysis. Conformity and homogeneity indices for the PTV-VOI were calculated by direct adaptation of the PTV-based formulations, with all volumetric and dose–volume parameters restricted to the PTV subvolume located within the VOI (Figure 2). Accordingly, only those portions of the reference isodose lying within the VOI were included in the calculation. This approach ensured methodological consistency between the global PTV evaluation and the localized analysis of the field junction region, while preserving the original mathematical definitions of the quality metrics. Underdosage phenomena were also observed in the caudal junction region (IC2); however, the analysis was restricted to IC1, as dose patterns in IC2 were comparable and further evaluation was not expected to yield additional insight.
Results
Dose coverage as a function of overlap field size. Figure 5 illustrates the D98% values for both the PTV and PTV-VOI across different overlap field sizes for each patient. For patient #1, the criterion D98% ≥95% from Figure 1 is achieved for the PTV with field overlap sizes of 6 cm or more, while for the PTV-VOI it is achieved with overlaps of 8 cm or more. For Patient #2, the D98% >95% is achieved for the PTV at all field overlap sizes, while the PTV-VOI meets the criterion from 6 cm onward. In Patient #3, the PTV satisfies the criterion across all field overlap lengths, whereas the PTV-VOI does it only for overlaps ≥8 cm. For Patient #4, D98%≥95% is fulfilled for both the PTV and the PTV-VOI only at overlap lengths of 10 cm or greater. For patient #5 the criterion is achieved at ≥ 8 cm for the PTV, and for the PTV-VOI from ≥10 cm. For patient #6 the D98% is achieved at all overlap sizes for the PTV, and for the PTV-VOI at ≥ 10 cm. With a field overlap size of 12 cm, the D98% for the PTV-VOI exceeds that of the PTV in all patients.
Dose coverage (D98%) values for the six patients [planning target volume (PTV) in blue and the subvolume (PTV-VOI) in orange] as a function of overlap field size.
Evaluation of conformity indices. Figure 6A-D shows the CI calculated according to the definitions in Figure 1 for the entire PTV (blue), as well as the VOI-based CI indices defined in Figure 2 for the PTV-VOI located in the field junction region. The CI are presented as a function of field overlap length for the four calculation methods RTOG, Salt–Lomax, Lomax, and Van’t Riet. For improved clarity, the CI are not shown on a patient-specific basis but are averaged across all patients. The VOI-based CI were normalized to 1.0. Due to their definition, these indices do not represent absolute CI. However, this approach allows a clearer visualization and a direct comparison of relative trend developments between the different indices and overlap lengths. Across all conformity indices, the global PTV-based values remained largely constant over the entire range of investigated field overlap lengths. The RTOG, Salt–Lomax, Lomax, and Van’t Riet conformity indices for the PTV consistently showed values close to 1.0 and exhibited no systematic dependence on overlap width. In contrast, the VOI-based conformity indices for the PTV-VOI showed a pronounced dependence on field overlap length. For the RTOG and Salt–Lomax indices, the conformity of the PTV-VOI increased continuously with increasing overlap length. The Lomax conformity index exhibited a different behavior. While the PTV-based Lomax index remained high and stable across all overlap lengths, the VOI-based Lomax index for the PTV-VOI showed a slight decrease with increasing overlap width. For the Van’t Riet conformity index, the PTV-based values remained almost at 1.0, and the PTV-VOI values showed only minimal variations across all overlap lengths without a systematic trend as a function of overlap length. Overall, the comparison between global PTV-based and localized PTV-VOI-based conformity indices shows that changes in field overlap length have little effect on global conformity metrics, whereas locally restricted conformity indices exhibit clear and systematic trend patterns in the field junction region.
Mean conformity indices (CI), averaged across all patients, for the planning target volume (PTV, blue) and the junction-focused subvolume (PTV-VOI, orange) as a function of field overlap length. Panels show (A) RTOG CI, (B) Salt–Lomax CI, (C) Lomax CI, and (D) Van’t Riet CI. For clarity, PTV-VOI-based values were normalized to 1.0; therefore, these curves illustrate relative trends across overlap lengths rather than absolute CI values.
PTV homogeneity. Figure 7 below shows the HI, averaged across all patients, as a function of field overlap (cm) for the PTV and the PTV-VOI. For both PTV and PTV-VOI, the HI value decreases with increasing field overlap. The PTV-VOI consistently shows slightly lower HI values than PTV. The largest difference between the two curves occurs at approximately 8 cm overlap. From 8 cm overlap onwards, both curves flatten out. Figure 4A-E shows enlarged sections of the dose distribution in the field transition area using the example of patient #1. The dose is shown in colour wash representation of 95%-Dmax of the prescribed dose. With an overlap of 4 cm, the dose distribution still shows considerable dose gaps. As the field overlap increases, the dose gaps close and the dose distribution becomes more homogeneous. With an overlap of 12 cm, the dose gaps are closed and the dose distribution is homogeneous.
Homogeneity index (HI) as a function of field overlap width, averaged across all patients, shown separately for the planning target volume (PTV, blue) and the junction-focused subvolume (PTV-VOI, orange).
Discussion
In CSI, VMAT offers significant advantages compared to 3D-CRT or IMRT techniques in terms of dose conformity and sparing of organs at risk (1-3, 6, 7). Achieving a homogeneous dose distribution that encompasses the entire PTV at the field transitions remains one of the greatest technical challenges (4, 10, 13). The aim of this study was to systematically investigate and quantitatively evaluate the influence of different field overlap lengths in VMAT-based CSI on the quality of dose coverage in the field transition zone. For this purpose, the D98% coverage, conformity indices CI according to RTOG, Lomax, Salt-Lomax, and Van’t Riet, as well as the homogeneity index HI according to ICRU 62 (17) were calculated. A central aspect of the study concerns the question of the extent to which global PTV-based quality metrics are sufficiently sensitive to detect locally limited dose deficits, especially in the sensitive field overlap areas. Due to the large irradiation volume of a craniospinal axis (CSA), small-scale underdoses in these transition areas may be statistically attenuated by averaging effects and being undetected, even though they are clinically relevant for various reasons: The CSA encompasses the entire cerebrospinal fluid system. Locally limited underdosing at field transitions can potentially serve as a nidus for tumor cell colonization and recurrence (19). In addition, a dose-dependent effect on local tumor control has been described for many central nervous system tumors. Failure to meet defined coverage criteria (e.g., D95% /D98%) can therefore impair local control (20). The field junction regions are particularly sensitive to setup and geometric errors (21). Strategies such as field feathering were developed precisely to minimize such hot and cold spots (22). Therefore, locally focused Quality Assurance (QA) and robustness analyses as well as a targeted assessment of junction dosimetry are necessary (14).
For these reasons, analyses were performed both for the entire PTV and for a defined PTV subregion (PTV-VOI), which was geometrically specified using a separately delineated volume of interest (VOI) positioned in the field overlap region. This approach enabled a differentiated evaluation of the transition region and facilitated the detection of potential small-scale dose deviations.
These theoretical considerations are confirmed by the results of the present investigation. In all six patients, increasing field overlap resulted in a continuous improvement in D98% values for both the PTV and the PTV-VOI. While the PTV met the acceptance criterion of D98% ≥ 95% in three out of six cases already at an overlap length of 4 cm, the PTV-VOI demonstrated markedly lower robustness. For the PTV-VOI, only one case fulfilled the acceptance criterion at an overlap of 6 cm, and three cases at 8 cm. From an overlap length of 10 cm onward, all cases met the D98% ≥ 95% criterion, and at 12 cm the D98% values of the PTV-VOI exceeded those of the PTV. This clearly shows that dose coverage in the field transition area depends much more on the overlap width than global PTV metrics would suggest. This supports our thesis that an exclusive evaluation at the PTV level obscures local underdosing in the transition area. To ensure sufficient dose coverage and compliance with the D98% acceptance criterion in the field transition area, an overlap length of at least 10-12 cm appears to be necessary. Similar observations have also been described in earlier studies, in which increased overlap led to improved dose distribution and lower gradients in the junction zones (12-14).
The evaluation of the CI also shows a differentiated picture between global, PTV-based, and local, PTV-VOI based CI values. While the PTV CIs for all four calculation methods remained very high at values around 1.0 across the entire range of field overlaps, the PTV-VOI CIs for RTOG, Salt-Lomax, and Lomax showed a clear dependence on the overlap width. For RTOG and SALT-Lomax, the PTV-VOI CI increased with increasing field overlap (RTOG 0.85-1.0, SALT-Lomax 0.92-1.0). In contrast, the Lomax method showed the opposite trend: the PTV-VOI CI decreased slightly with increasing overlap (from approximately 1.0 to 0.9). Similar to the PTV-CIs, the PTV-VOI CIs according to Van’t Riet remained almost constant at 1.0.
The definitions of the conformity indices shown in Figure 1 illustrate that the evaluation of dose distribution focuses on different aspects depending on the mathematical formulation. The RTOG measures the ratio of the PTV and the volume of the reference isodose. Salt–Lomax primarily evaluates the coverage of the target volume by the prescribed isodose; the Lomax index also responds to local dose excesses in normal tissue. The Van’t Riet index considers the dose within and also outside the target volume (18). The RTOG CI describes the ratio between the volume covered by the prescription isodose (VRI) and the target volume (TV). An increase formally indicates a numerical alignment of the volumes in size, but not in position. This upward trend is very clear in the PTV-VOI CI values of the calculations. However, the RTOG-CI does not allow any conclusions to be drawn about the spatial location or shape accuracy of the volumes. The smoothing of the dose distribution visible in Figure 4A-E confirms that the VRI/TV ratio becomes more favorable as the overlap width increases. However, the images also show that the volumes increasingly overlap, thereby stabilizing the dose distribution in the transition area and reducing dose gaps. The global PTV metric for RTOG remains largely unaffected by the varying field length. This indicates a statistical leveling of local underdoses when considering the total volume exclusively. The SALT–Lomax CI exclusively evaluates the proportion of the target volume covered by the reference isodose (TVRI) and thus provides information about underdoses within the PTV. However, it remains insensitive to a possible extension of the reference isodose beyond the target volume into normal tissue. A value of 1, for example, can also be achieved when normal tissue is irradiated, regardless of the size of its volume. Our evaluation shows an increase in SALT-Lomax PTV-VOI CI with increasing field overlap length. As with the RTOG, this is reflected in Figure 4A-E by increasing dose coverage. Although exceeding the reference isodose beyond the PTV is a noticeable trend in Figure 4A-E, this is not detected by the SALT-Lomax CI as described above and is not considered clinically relevant overdose in the given cases. As with the RTOG, the global PTV metric remains largely unaffected by the field lengths. The Lomax CI (TVRI/VRI) describes the ratio between the proportion of the target volume covered by the reference isodose (TVRI) and the total volume of the reference isodose (VRI). It is therefore a geometric measure of the spatial precision of the dose distribution and indicates how selectively the radiation covers the target volume. A value of 1 indicates perfect spatial agreement between the reference isodose and the target volume – the irradiated area covers the target volume exactly without overdosing the surrounding tissue. Decreasing values indicate that an increasing proportion of the reference isodose lies outside the target volume, i.e., adjacent normal tissue is being over-irradiated. The index therefore does not provide direct information about target coverage, but rather evaluates the spatial precision of dose coverage. The results of this study showed that the Lomax index for the PTV was stable and close to 1 across all field overlap sizes, indicating consistent geometric coverage of the target volume. For the PTV-VOI, the Lomax CI decreased slightly with increasing field overlap – from approximately 1.0 at 4 cm to approximately 0.9 at 12 cm. This means that the irradiated isodose region (VRI) expanded slightly beyond the defined PTV-VOI area as the overlap width increased. This effect can be seen in Figure 4A-E: as overlap increases, the dose distribution becomes more homogeneous, but the 95% isodose spreads slightly beyond the original field boundaries. The slight decrease in PTV-VOI Lomax CI with larger overlaps therefore does not reflect a deterioration in plan quality, but rather an increasing smoothing and spatial expansion of the dose distribution in the field transition region as a result of improved dose coverage. The Van’t Riet CI represents a combined assessment of target coverage and dose excess. It considers both the proportion of the target volume covered by the reference isodose (TVRI/TV) and the proportion of the reference isodose outside the target volume (TVRI/VRI). An ideal value of 1.0 is achieved when the reference isodose completely covers the target volume without extending into the surrounding tissue. The Van’t Riet Index thus integrates the advantages of the Salt–Lomax and Lomax approaches and is considered a particularly robust measure for evaluating dose precision. In the present study, the Van’t Riet indices for both the PTV and the PTV-VOI remained almost constant at values around 1.0 across all field overlap widths, with the PTV-VOI values always slightly below the PTV values.
However, the moderate expansion of the irradiated isodoses beyond the target volume, as shown in Figure 4A-E, is hardly captured by the Van’t Riet index. This is understandable, as the index, due to its definition – in particular the quadratic weighting of the covered target volume – can only react to a limited extent to locally limited dose phenomena. Although the slightly lower PTV-VOI values indicate a minimal, locally limited extension of the isodoses in the transition areas, this is hardly reflected in the quadratic weighting of the index. For the cases presented here, the Van’t Riet index therefore proved to be unsuitable for precisely characterizing small-scale deviations in dose distribution in the field transition area. We were able to show that the initial hypothesis that global PTV-based quality metrics can mask locally limited dose deficits is true in the present cases. In conclusion, this means that clinics should not rely exclusively on conventional and familiar global quality metrics when planning a CSI, as these do not have the necessary sensitivity to reliably reflect critical dose fluctuations, especially in field overlap areas. It is therefore advisable to develop your own quality criteria, as is the case in this study.
Limitations. Only four patient data sets were available to support the hypothesis. In addition, to ensure the comparability of the four assessment methods (RTOG, Salt–Lomax, Lomax, and Van’t Riet), the respective CI values were averaged across patients and then normalized to 1.0. This facilitates the visualization of trends, but does not mean that an absolute CI value of 1.0 was actually achieved in the individual data sets. Nevertheless, due to the careful data preparation and consistent trends, we consider the results to be statistically recognizable and clinically interpretable.
This work is theoretically oriented, but addresses a topic that will become increasingly relevant in the future: the growing use of AI-based methods in everyday clinical practice, particularly in the field of automated or AI-supported optimization of radiation therapy plans. The findings can help not only to better interpret dose distributions, but also to control them in a targeted manner in the future. In addition to the analysis of D98%, the combination of the RTOG, Salt–Lomax, and Lomax conformity indices in particular proved to be a suitable tool for the systematic evaluation and potential algorithmic control of optimization processes. While the RTOG index allows conclusions to be drawn about the volumetric ratios of the reference isodose with the target volume, the Salt–Lomax index provides information about the geometric position of the target volume coverage. The Lomax index complements this perspective by providing information about the extent to which the reference isodose diverges into the surrounding normal tissue.
Conclusion
The initial hypothesis that locally limited dose deficits in the field transition areas are statistically leveled out by global, PTV-based quality metrics and thus inadequately represented was clearly confirmed by the results of this study. While the global PTV metrics remained largely constant across all field overlap lengths examined, the PTV-VOI based metrics – in particular D98% and the PTV-VOI CIs according to RTOG, Salt–Lomax, and Lomax – responded much more sensitively to dose gaps and spatial deviations in the transition area. The PTV-VOI analysis repeatedly showed reduced dose precision for small overlaps, which remained completely hidden in the global PTV values. Thus, local, PTV-VOI based analysis proved to be diagnostically superior when it comes to detecting subtle, clinically relevant deviations in dose distribution at the field junctions.
Our results also show that a field overlap of at least 10-12 cm is necessary to ensure robust and clinically acceptable dose coverage in the transition area. The combination of D98% analysis and the structurally complementary conformity indices RTOG, Salt–Lomax and Lomax have proven particularly suitable for mapping both volumetric and geometric precision aspects of dose distribution in the junction area. Against this background, clinics should critically question the sole use of established global quality metrics in CSI planning. Local, structure-specific metrics are required for a reliable assessment of plan quality, especially in technically and clinically sensitive field transition regions. The results of this study therefore underscore the need to systematically integrate PTV-VOI based dose metrics into the routine evaluation of CSI plans in order to detect critical dose deviations at an early stage and improve the long-term robustness of treatment.
With the foreseeable increase in the use of AI-based optimization and quality assurance algorithms in radiation therapy, the analysis presented here takes on additional significance. AI systems benefit particularly from clearly defined, sensitive, and locally effective quality metrics. The present results can therefore serve as a basis for specifically aligning AI-supported optimization processes with junction dosimetry and ensuring that critical local dose deviations are not “masked” by global metrics. This will enable future AI methods not only to interpret dose distributions, but also to actively control and stabilize them.
Footnotes
Authors’ Contributions
The Authors participated in the design of the study. The data were collected by C.Z. and interpreted by C.Z., F.C., L.M., and C.H. The article was drafted by C.Z. and A.L., and initially reviewed by D.R. The final version was reviewed and approved by the Authors.
Conflicts of Interest
The Authors indicate that there are no conflicts of interest related to this study.
Artificial Intelligence (AI) Disclosure
No artificial intelligence (AI) tools, including large language models or machine learning software, were used in the preparation, analysis, or presentation of this manuscript.
- Received February 9, 2026.
- Revision received March 3, 2026.
- Accepted March 6, 2026.
- Copyright © 2026 The Author(s). Published by the International Institute of Anticancer Research.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.