📄 medicine

Dosimetric Impact of CT/MRI-Based Target Delineation Differences in Radiotherapy for Brain Metastases Close to the Skull

This study demonstrates that relying solely on CT for target delineation in brain metastases near the skull leads to significant underestimation of tumor volume and inadequate dose coverage compared to MRI-based delineation, particularly in local and simultaneous integrated boost radiotherapy, thereby highlighting the necessity of integrating contrast-enhanced MRI into treatment planning to prevent local failure.

Original authors: Song Sun, Guanzhong Gong, Lu Zhao, Yong Yin

Published 2026-07-17

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Song Sun, Guanzhong Gong, Lu Zhao, Yong Yin

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are a master archer trying to hit a tiny, moving target hidden inside a dense, foggy forest. In the world of cancer treatment, that target is a tumor, and the "fog" is the human body. To hit the target with a laser beam of radiation, doctors need a perfect map. Usually, they use two types of maps: a standard CT scan, which is like a black-and-white sketch good for seeing bones but a bit blurry on soft tissues, and an MRI, which is like a high-definition, colorful photograph that shows the soft details of the brain much better. When doctors plan radiation, they draw a circle around the tumor on these maps and then add a little extra "safety buffer" around it to make sure they don't miss. This is called the Planning Target Volume. The big question is: if you draw that circle based on the blurry sketch versus the high-definition photo, does it actually matter? Does the difference in the drawing change how much "sunlight" (radiation dose) actually hits the target? This is especially tricky when the tumor is hiding right next to the skull, where the bone can create weird shadows and distortions on the standard map.

This study, led by researchers at Shandong First Medical University and Shandong Cancer Hospital, decided to test exactly that. They looked at 245 brain tumors that were sitting right up against the skull (a group they call "brain metastases close to the skull"). They took the same patients and drew the tumor boundaries three different ways: once using the standard "brain-window" CT (the blurry sketch), once using a "bone-window" CT (a different setting on the sketch to see the bone better), and once using the high-definition MRI. Then, they took the actual radiation plans the patients had already received and checked what happened when they applied those plans to the different drawings.

Here is the twist they found: the MRI consistently showed the tumors were bigger than the CT scans did. It's like the MRI saw a few extra inches of the target that the CT sketch missed, especially right where the tumor touched the skull. When the researchers checked the radiation dose on these "MRI-only" extra bits, they found a problem. Because the original plans were drawn based on the smaller CT sketches, those extra bits of tumor often ended up in the "shadow" of the radiation beam, receiving less dose than intended.

The study measured this using a specific metric called D98%, which tells you how much dose the lowest-dosed 98% of the tumor receives. When they used the MRI-based maps, the dose dropped significantly compared to the CT-based maps. In the group of patients receiving a "Simultaneous Integrated Boost" (a very precise, high-dose treatment), the dose dropped by as much as 3.16 Gy (a unit of radiation) compared to the CT plan. In fact, for these high-dose treatments, nearly 62% of the tumors had parts that received less than the full prescribed dose when viewed through the MRI lens. Even worse, about 31% of these tumors had parts that received less than 95% of the dose, which the researchers call a "clinically relevant undercoverage."

The researchers suggest that this happens because the extra tumor parts seen on the MRI are usually located right on the edge of the radiation beam, where the dose falls off quickly. If you draw your target too small (using only the CT), you might think you are hitting the bullseye, but the MRI reveals that the edge of the target is actually slipping into the low-dose zone. This effect was most dangerous in the high-precision treatments (LRT and SIB) where the dose drops off sharply, but it was less of an issue in the whole-brain treatments where the dose is more spread out.

The paper argues against the idea that simply adding a bigger safety buffer to the CT plan is the solution. Instead, they suggest that for tumors touching the skull, doctors should always use the high-definition MRI fused with the CT scan to draw their targets. They found that relying only on the CT scan can lead to "occult undercoverage," a fancy way of saying the plan looks perfect on the CT map, but the MRI map shows the tumor is actually being under-dosed. The study concludes that for these specific, skull-hugging tumors, especially those getting high-dose treatment, using the MRI to guide the drawing is essential to ensure the entire tumor gets the full punch of the radiation.

Technical Summary: Dosimetric Impact of CT/MRI-Based Target Delineation Differences in Radiotherapy for Brain Metastases Close to the Skull

Problem Statement
Accurate target delineation is critical for high-precision radiotherapy of brain metastases (BMs), particularly those located close to the skull (BMCS). While planning CT is the standard for dose calculation, its soft-tissue contrast is limited, and BMCS are susceptible to beam-hardening artifacts and partial-volume effects near the inner table of the skull. Although contrast-enhanced MRI is the gold standard for defining tumor boundaries, it is often not the primary dataset for dose calculation. The extent to which delineation differences between brain-window CT, bone-window CT, and contrast-enhanced MRI translate into clinically significant dosimetric undercoverage—specifically in high-dose gradient techniques like Local Radiation Therapy (LRT) and Simultaneous Integrated Boost (SIB)—remains insufficiently quantified.

Methodology
This retrospective study analyzed 245 BMCS lesions from 132 patients treated with Whole-Brain Radiation Therapy (WBRT), LRT, or SIB. The study utilized a "dose-mapping" approach without reoptimizing treatment plans to isolate the dosimetric impact of delineation geometry.

Image Acquisition: Patients underwent contrast-enhanced CT (brain-window and bone-window settings) and contrast-enhanced T1-weighted MRI simulation within 48 hours. Images were rigidly registered using MIM Maestro software.
Target Delineation: Gross Tumor Volumes (GTVs) were manually delineated on three datasets: brain-window CT ( $GTV_{Brain}$ ), bone-window CT ( $GTV_{Skull}$ ), and contrast-enhanced MRI ( $GTV_{MRI}$ ). A uniform 3-mm expansion was applied to generate corresponding Planning Target Volumes ( $PTV_{Brain}$ , $PTV_{Skull}$ , $PTV_{MRI}$ ).
Region Definition: "MRI-only" regions were defined via Boolean operations as the portions of $PTV_{MRI}$ extending beyond $PTV_{Brain}$ ( $PTV_{MRI-Brain}$ ) or $PTV_{Skull}$ ( $PTV_{MRI-Skull}$ ).
Dosimetric Analysis: The original clinical dose distributions were mapped onto all PTV definitions. Key metrics included $D_{98\%}$ (dose to 98% of the volume), $D_{2\%}$ , $D_{mean}$ , and Homogeneity Index (HI).
Event Classification: The study defined "below-prescription-dose events" ( $nD_{98\%} < 1.00$ ) and "clinically relevant undercoverage" ( $nD_{98\%} < 0.95$ ). "Occult undercoverage" was identified when CT-based targets met coverage criteria ( $nD_{98\%} \geq 0.95$ ) but the MRI-referenced target or MRI-only regions did not.
Statistics: Non-parametric tests (Friedman, Wilcoxon signed-rank) and correlation analyses were used to assess differences and relationships between volume reduction and dose deviation.

Key Contributions and Results

Systematic Underestimation by CT: Across all treatment modalities, $PTV_{MRI}$ was consistently larger than both $PTV_{Skull}$ and $PTV_{Brain}$ ( $P < 0.001$ ). $PTV_{Brain}$ showed the greatest volume reduction relative to $PTV_{MRI}$ , followed by $PTV_{Skull}$ .
Dosimetric Impact on Marginal Coverage: The geometric discrepancies resulted in significant reductions in $D_{98\%}$ $D_{98%}$ for MRI-based targets compared to CT-based targets.
- WBRT: $D_{98\%}$ reductions were 0.68 Gy ( $PTV_{Brain}$ ) and 0.58 Gy ( $PTV_{Skull}$ ).
- LRT: Reductions increased to 1.96 Gy and 1.90 Gy, respectively.
- SIB: Reductions were most pronounced at 3.16 Gy and 2.82 Gy, respectively.
Location of MRI-Only Regions: MRI-only regions were predominantly located in low-dose areas near the prescription isodose line. Consequently, these regions exhibited lower $D_{98\%}$ and $D_{mean}$ values compared to the overall $PTV_{MRI}$ .
Incidence of Undercoverage:
- The incidence of clinically relevant undercoverage ( $nD_{98\%} < 0.95$ ) for $PTV_{MRI}$ was 6.82% (WBRT), 17.91% (LRT), and 31.11% (SIB).
- In the SIB group, clinically relevant undercoverage in MRI-only regions reached 41.11% ( $PTV_{MRI-Brain}$ ) and 42.22% ( $PTV_{MRI-Skull}$ ).
Occult Undercoverage: A significant proportion of lesions showed "occult undercoverage," where CT-based evaluation suggested adequate coverage ( $nD_{98\%} \geq 0.95$ ), but the MRI-referenced target or MRI-only regions failed to meet this threshold. This risk was highest in the SIB group.
Correlation with Treatment Modality: In LRT and SIB groups, a significant positive correlation was found between the percentage volume reduction of CT-delineated targets and the deviation in $nD_{98\%}$ . No such correlation was observed in WBRT, likely due to its more homogeneous dose distribution.
Risk Stratification: Lesions with a high fraction of MRI-only volume had a 2.05-fold higher risk of clinically relevant undercoverage in LRT and SIB compared to those with low fractions.

Significance and Claims
The paper concludes that brain-window and bone-window CT-based delineation systematically underestimates the target extent of BMCS compared to contrast-enhanced MRI. This geometric discrepancy translates directly into reduced marginal dose coverage, particularly in high-dose gradient treatments (LRT and SIB).

The authors assert that for BMCS treated with LRT or SIB, relying solely on CT-based target evaluation may mask clinically relevant undercoverage in the MRI-defined target or MRI-only regions. Therefore, the study recommends that GTV delineation for these lesions should routinely incorporate registered contrast-enhanced MRI alongside planning CT. While bone-window CT can assist in assessing the skull interface, it should not replace contrast-enhanced MRI for defining soft-tissue tumor boundaries. The authors emphasize that indiscriminate margin expansion is not a substitute for accurate boundary identification; rather, rigorous fusion and review of MRI data are necessary to mitigate the risk of local failure.

The study acknowledges limitations, including its retrospective nature, the lack of plan reoptimization, and the absence of recurrence correlation, noting that future work should validate these dosimetric findings with clinical outcomes.

Technical Summary: Dosimetric Impact of CT/MRI-Based Target Delineation Differences in Radiotherapy for Brain Metastases Close to the Skull

More like this