Ex) Article Title, Author, Keywords
Ex) Article Title, Author, Keywords
Progress in Medical Physics 2024; 35(4): 145-154
Published online December 31, 2024
https://doi.org/10.14316/pmp.2024.35.4.145
Copyright © Korean Society of Medical Physics.
Sohyun Ahn1,2 , So Eun Choi3
, Jeong-Heon Kim4,5,6
, Kwangwoo Park7
, Hai-Jeon Yoon8
Correspondence to:Sohyun Ahn
(mpsohyun@gmail.com)
Tel: 82-2-6986-6305
Fax: 82-0504-158-4052
This is an Open-Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Purpose: This study investigated the use of synthetic computed tomography (CT) images derived from cone beam CT (CBCT) scans to analyze dose changes in breast cancer patients undergoing treatment and to evaluate the optimal timing for implementing adaptive radiotherapy.
Methods: A retrospective analysis was conducted on five breast cancer patients treated with tomotherapy-based volumetric-modulated arc therapy at Yongin Severance Hospital. Each patient received 15 fractions, with doses of 320 centigray (cGy) to the high-dose planning target volume (PTV) and 267 cGy to the low-dose PTV. Planning CT images were acquired using the Aquilion scanner, and CBCT images were captured with the VersaHD linear accelerator’s on-board imager. These images were registered in RayStation using a hybrid deformable image registration method to generate synthetic CT images. Dose distributions were reanalyzed using the synthetic CT images, and dosevolume histogram parameters, including the dose to 95% of the volume (D95) and mean dose (Dmean) for the PTV, as well as D95, Dmean, the percentage of the volume receiving at least 5 Gy (V5) and 10 Gy (V10) for organs-at-risk (OARs), were extracted using MATLAB to assess dose changes during treatment.
Results: For the original plans, the mean D95 for PTV high across all patients was 287.13±31.32 cGy, while for PTV low, it was 245.53±6.21 cGy. In contrast, the adaptive plans yielded a mean D95 of 298.17±12.37 cGy for PTV High and 247.25±4.23 cGy for PTV low. The ART Plan may lead to increased dose exposure in certain structures, such as the spinal cord, while providing targeted improvements in reducing radiation exposure in specific OARs (e.g., contralateral breast and esophagus).
Conclusions: Synthetic CT images generated from CBCT scans provide a fast and efficient means of quantifying dose changes, supporting precise patient care through interfractional evaluation. Future studies will aim to apply this method to other organs and larger patient cohorts.
KeywordsSynthetic computed tomography generation, Adaptive radiotherapy, Breast cancer treatment, Dose-volumetric analysis, Cone beam computed tomography imaging
Breast cancer remains one of the most prevalent malignancies affecting women globally, with high mortality rates despite advancements in early detection and treatment technologies. Radiation therapy plays a pivotal role in breast cancer management, particularly as an adjuvant therapy following surgery to reduce the risk of tumor recurrence and improve survival outcomes [1-3]. Traditional radiation therapy planning is typically based on computed tomography (CT) images obtained during the initial treatment stage. However, anatomical changes or tumor shifts occurring during therapy can compromise the accuracy of the initial treatment plan [3-5].
To mitigate this challenge, adaptive radiation therapy (ART) has been introduced. ART adapts the treatment plan in response to anatomical changes over time, thereby allowing for more precise and effective radiation delivery. Despite its advantages, ART requires multiple CT scans, which can result in additional radiation exposure and increased time and cost burdens for patients.
A promising solution to this problem is the use of synthetic CT derived from cone-beam CT (CBCT). Synthetic CT provides images of similar quality to conventional CT without additional radiation exposure, thereby enabling accurate assessment of anatomical changes in patients. This approach enhances the efficiency of ART and allows for treatment plan optimization [5-7].
The primary aim of this study was to generate synthetic CT images from CBCT images and use them to evaluate treatment accuracy in patients with breast cancer undergoing ART. This approach allows timely adjustments to treatment plans, thereby maximizing therapeutic efficacy while minimizing side effects. Future research will focus on expanding the applicability of this method to other anatomical regions, further demonstrating the robustness and versatility of synthetic CT-based ART evaluation [8,9].
This retrospective study included five female patients with breast cancer who were treated at the Yongin Severance Hospital between March 2020 and July 2023. The patients were treated using tangential volumetric modulated arc therapy (T-VMAT) at doses of 320 centigray (cGy) for planning target volumes (PTV) high and 267 cGy for PTV low, over 15 fractionated sessions. Informed consent was obtained from all participants (Table 1).
Table 1 Patient characteristics
Variable | Value |
---|---|
Sex (n) | |
Male | |
Female | 5 |
Age (y) | |
Mean (range) | 53 (40–68) |
Median | 52 |
Fraction, mean | |
Total fractions | 15 |
Number of ARTs (range) | 11 (8–13) |
Planning CT images were acquired using an Aquilion CT scanner (Canon Medical Systems). The patients were placed in the supine position with their arms raised above their heads to simulate the treatment conditions. The scans were performed with a slice thickness of 3 mm to provide the detailed anatomical information necessary for precise treatment planning. CBCT images were obtained using the On-Board Imager system integrated with a VersaHD LINAC (Elekta). These scans were performed before each treatment session to monitor anatomical changes and ensure accurate patient positioning. The CBCT imaging protocol included a full gantry rotation to capture comprehensive volumetric data, which were then used to generate synthetic CT images (Fig. 1).
Synthetic CT images were generated from CBCT scans using a hybrid deformable image registration method within the RayStation treatment planning system (RaySearch Laboratories). This approach involved the following steps:
1) Rigid registration: Initial alignment of CBCT images to planning CT based on bony anatomy.
2) Deformable registration: Fine-tuning the alignment to account for soft tissue variations and patient positioning differences. This process ensures that the critical structures and target volumes are accurately represented.
3) Quality assurance: The generated synthetic CT images were reviewed for accuracy in anatomical representation and were validated against planning CT images.
All treatment plans were developed using RayStation. The following steps were undertaken:
1) Contouring: Delineation of the PTVs and organs-at-risk (OARs) in the planning CT images. The OARs included the breast, heart, ipsilateral lung, spinal cord, esophagus, and contralateral breast.
2) Plan optimization: Optimization of the T-VMAT plans to achieve adequate dose coverage of the PTV while minimizing exposure to OARs. The dose prescriptions were set at 320 cGy for PTV high and 267 cGy for PTV low, delivered over 15 fractions.
3) Verification: Each treatment plan was reviewed and adjusted based on synthetic CT images to account for anatomical changes throughout the treatment course.
Dosimetry parameters were extracted and compared between the original and adaptive treatment plans through synthetic CT-based evaluation using dose–volume histogram (DVH) metrics. The following parameters were analyzed: D95, dose received by 95% of the target volume; Dmean, mean dose received by the target volume; and V5 and V10, volumes of the OARs receiving 5 and 10 Gy, respectively.
The dataset included paired data for each patient, comprising the Original and ART Plans. On average, ART data covered 11 fractions per patient. To robustly analyze the dose differences between the two plans, a linear mixed model (LMM) was applied.
The primary goal of employing LMM was to evaluate the differences between the Original and ART Plans with respect to DVH parameters in the same region of interest.
The LMM used in this study was structured as follows:
(1) Dependent variable: Log-transformed DVH values were selected as the dependent variable to normalize the data distribution and address variability in the ranges.
(2) Independent variable: The treatment plan, which was categorized as either “Original Plan” or “ART Plan,” was modeled as the fixed effect to capture the effects on DVH differences.
(3) Random effect: Patient ID was included as a random effect to adjust for individual patient-specific variability, allowing for generalizable analysis without conflating interpatient differences with treatment effects.
For each patient, the mean dose volume was calculated separately for the ART Plan and Original Plan:
(1) Original Plan mean per patient: The mean of each fraction’s DVH parameters in the Original Plan group.
(2) ART Plan mean per patient: The mean of each fraction’s DVH parameters in the ART Plan group.
(3) Average Original Plan: The sum of the Original Plan Means per Patient divided by the total number of patients.
(4) Average ART Plan: The sum of the ART Plan Means per Patient divided by the total number of patients.
For each patient, the difference between the ART Plan and Original Plan means is calculated as follows:
(1) Difference per patient: The Original Plan Mean per Patient is subtracted from the ART Plan Mean per Patient for each patient.
(2) Mean difference: The sum of the Difference per Patient divided by the total number of patients.
(3) Relative difference (%): The mean difference is divided by the average Original Plan and then multiplied by 100 to express the result as a percentage.
All procedures involving human participants were performed according to the ethical standards of the Institutional and/or National Research Committee and the 2013 Helsinki Declaration and its later amendments or comparable ethical standards.
Based on the synthetic CT-based dosimetry analysis, the D95 and Dmean for PTV high exhibited notable differences between the original and adaptive treatment plans. Table 2 presents the LMM results, and Table 3 presents the
Table 2 LMM results for PTV and OARs
ROI | Variable | Group coefficient (Ref. Original Plan) | Group standard error (Ref. Original Plan) | Group variance |
---|---|---|---|---|
PTV high | D95 | 0.043 | 0.010 | 0.006 |
Dmean | 0.013 | 0.003 | 0 | |
PTV low | D95 | 0.007 | 0.005 | 0 |
Dmean | 0.001 | 0.001 | 0 | |
Heart | V5 | −0.006 | 0.010 | 0.005 |
V10 | 0.005 | 0.004 | 0.003 | |
D95 | −0.007 | 0.008 | 0.005 | |
Dmean | 0.009 | 0.009 | 0.003 | |
Spinal cord | V5 | 0.002 | 0.004 | 0.117 |
V10 | 0.031 | 0.006 | 0.028 | |
D95 | 0.175 | 0.040 | 0.086 | |
Dmean | 0.023 | 0.020 | 0.348 | |
Esophagus | V5 | 0.046 | 0.022 | 0.138 |
V10 | −0.010 | 0.002 | 0.016 | |
D95 | 0.017 | 0.048 | 0 | |
Dmean | −0.023 | 0.018 | 0.171 | |
Ipsilateral lung | V5 | −0.004 | 0.006 | 0.047 |
V10 | 0.036 | 0.007 | 0.060 | |
D95 | 0.019 | 0.011 | 0.029 | |
Dmean | 0.017 | 0.009 | 0.110 | |
Contralateral breast | V5 | −0.014 | 0.009 | 0.012 |
V10 | 0.006 | 0.010 | 0.012 | |
D95 | −0.065 | 0.018 | 0.020 | |
Dmean | −0.002 | 0.008 | 0 |
Table 3 Overall average dose per fraction for PTV and OARs across patients
ROI | Variable | Original Plan | ART Plan |
---|---|---|---|
PTV high | D95 | 287.13±31.32 | 298.17±12.37 |
Dmean | 316.80±9.50 | 320.88±4.53 | |
PTV low | D95 | 245.53±6.21 | 247.25±4.23 |
Dmean | 274.93±5.69 | 275.20±4.19 | |
Heart | V5 | 0.56±0.14 | 0.55±0.11 |
V10 | 0.06±0.07 | 0.07±0.06 | |
D95 | 6.19±0.42 | 6.16±0.64 | |
Dmean | 17.16±1.55 | 17.27±0.72 | |
Spinal cord | V5 | 0.33±0.53 | 0.33±0.49 |
V10 | 0.12±0.17 | 0.16±0.23 | |
D95 | 0.28±0.33 | 0.67±0.81 | |
Dmean | 8.48±5.51 | 9.04±5.93 | |
Esophagus | V5 | 0.49±0.62 | 0.55±0.54 |
V10 | 0.09±0.15 | 0.07±0.14 | |
D95 | 3.88±1.71 | 3.79±1.04 | |
Dmean | 13.63±5.67 | 13.39±5.68 | |
Ipsilateral lung | V5 | 2.45±0.63 | 2.45±0.72 |
V10 | 1.41±0.57 | 1.49±0.54 | |
D95 | 6.83±1.24 | 7.03±1.50 | |
Dmean | 45.83±14.08 | 46.86±14.77 | |
Contralateral breast | V5 | 0.82±0.26 | 0.79±0.15 |
V10 | 0.16±0.17 | 0.16±0.11 | |
D95 | 6.98±1.47 | 6.49±1.06 | |
Dmean | 18.60±1.28 | 18.53±0.94 |
Patient-specific dose distribution differences were also analyzed. Patient 1, who exhibited the smallest deviation between the original and adaptive treatment plans, exhibited relatively stable D95 values across both PTV high and PTV low. In contrast, Patient 4 had the largest difference between the two treatment plans, with significant variation in D95 values, reflecting the need for patient-specific consideration when performing ART. Fig. 3 presents the comparison of D95 and Dmean values for high PTV in Patients 1 and 4, emphasizing the role of synthetic CT-based plan evaluation in accommodating patient-specific differences.
The analysis of PTV coverage revealed significant improvements in dose conformity with the ART Plan compared with the Original Plan (Table 4). In particular,
Table 4 Relative differences between the Original Plan and ART Plan for PTV and OARs
ROI | Variable | Mean difference | Relative difference (%) | |
---|---|---|---|---|
PTV high | D95 | 11.04±19.38 | 3.85 | <0.001 |
Dmean | 4.08±8.45 | 1.29 | <0.001 | |
PTV low | D95 | 1.72±4.95 | 0.70 | 0.217 |
Dmean | 0.27±1.72 | 0.10 | 0.169 | |
Heart | V5 | −0.01±0.14 | −2.32 | 0.525 |
V10 | 0.01±0.03 | 7.82 | 0.142 | |
D95 | −0.03±0.47 | −0.47 | 0.370 | |
Dmean | 0.12±1.30 | 0.68 | 0.316 | |
Spinal cord | V5 | −0.01±0.05 | −1.74 | 0.614 |
V10 | 0.04±0.07 | 35.72 | <0.001 | |
D95 | 0.39±0.70 | 136.00 | <0.001 | |
Dmean | 0.55±1.41 | 6.50 | 0.271 | |
Esophagus | V5 | 0.06±0.30 | 12.04 | 0.042 |
V10 | −0.01±0.02 | −14.56 | <0.001 | |
D95 | −0.09±1.19 | −2.29 | 0.729 | |
Dmean | −0.24±2.40 | −1.78 | 0.195 | |
Ipsilateral lung | V5 | 0.00±0.20 | 0.18 | 0.535 |
V10 | 0.08±0.13 | 5.73 | <0.001 | |
D95 | 0.19±0.75 | 2.80 | 0.082 | |
Dmean | 1.03±4.06 | 2.25 | 0.067 | |
Contralateral breast | V5 | −0.04±0.16 | −4.43 | 0.102 |
V10 | 0.00±0.13 | 1.25 | 0.567 | |
D95 | −0.49±1.19 | −7.02 | <0.001 | |
Dmean | −0.07±1.66 | −0.37 | 0.813 |
(1) D95: The ART Plan demonstrated a significantly higher dose difference in PTV high (average 3.85%,
(2) Dmean: For PTV high, the ART Plan achieved a marginal reduction in Dmean, with a statistically significant difference (
These findings suggest that the ART Plan significantly enhances dose conformity in PTV high, thereby ensuring robust target coverage while maintaining precision in dose delivery.
We conducted a dosimetry analysis of the OAR structures, particularly the heart and ipsilateral lung, using synthetic CT-based evaluation across the cohort (Fig. 4 and Table 4).
(1) V5: ART exhibited minor improvements in reducing dose exposure for most OARs; however, these changes were not statistically significant. In contrast, the esophagus exhibited a significant dose increase (12.04%); this change was statistically significant (
(2) V10: ART achieved significant reductions in dose exposure for the esophagus (−14.56%). In contrast, significant dose increases were observed in the spinal cord (35.72%) and ipsilateral lung (5.73%) (
(3) D95: ART exhibited a significant reduction in dose exposure for the contralateral breast (average −7.02%,
(4) Dmean: Minor dose increases were noted for the spinal cord and ipsilateral lung, whereas the other OARs exhibited minor dose decreases. None of the average differences were statistically significant.
These findings suggest that the ART Plan increases dose exposure in certain structures, such as the spinal cord, while providing targeted improvements in reducing radiation exposure in specific OARs (e.g., contralateral breast and esophagus).
This study demonstrated the efficacy of using synthetic CT images derived from CBCT scans to guide ART for patients with breast cancer. The relative differences in DVH parameters, such as the increase in D95 for the high-dose PTV region from 287.13±31.32 cGy in the Original Plan to 298.17±12.37 cGy in the adaptive plan, highlight the importance of synthetic CT-based plan evaluation in enhancing treatment accuracy and dose conformity.
Moreover, the trends observed in the OARs underscore the complex effects of adaptive planning. Although the ART Plan effectively reduced dose exposure in specific structures, it also increased exposure in other structures, particularly the spinal cord. This highlights the necessity of precise adaptive planning, supported by inter-fractional evaluation, to mitigate potential risks. These findings suggest that ART can optimize the balance between target coverage and organ sparing but requires careful monitoring of dose tradeoffs to critical structures.
Synthetic CT in ART offers several clinical advantages. First, it reduces the need for repeated CT scans, thereby minimizing additional radiation exposure and associated risks for patients. This approach also enhances workflow efficiency by providing rapid and reliable anatomical assessments, allowing for timely treatment plan adjustments [2,10].
The significant dosimetry differences observed suggest that synthetic CT-based evaluation can enhance the precision of radiation delivery, potentially enhancing tumor control and reducing side effects. The observed differences in DVH parameters between the original and adaptive treatment plans indicate that synthetic CT can effectively capture dose distribution changes, thereby ensuring consistent and accurate treatment delivery [1,4,11].
Despite these promising findings, this study has several limitations. The sample size of five patients was relatively small, which may limit the generalizability of the results. Furthermore, the study was retrospective, which could introduce selection bias. Future studies should include larger, prospective cohorts to validate these findings and further investigate the benefits of synthetic CT-based plan evaluation in ART [3,12].
Another limitation is the potential for variability in synthetic CT generation due to differences in CBCT image quality and patient positioning. Standardizing the imaging and registration protocols could help reduce this variability and ensure more consistent results across different clinical settings [6,13].
Future research should focus on expanding the application of synthetic CT in ART to other anatomical regions and cancer types. Investigating the long-term clinical outcomes of patients treated with adaptive plan evaluation based on synthetic CT will provide valuable insights into the efficacy and safety of this approach. Furthermore, integrating advanced imaging techniques and machine learning algorithms can enhance the accuracy and efficiency of synthetic CT generation and dose prediction [2,10,11].
The development of robust, standardized protocols for synthetic CT generation and validation is crucial for widespread clinical adoption. Collaborative efforts between research institutions and clinical centers can facilitate the sharing of best practices and contribute to the development of comprehensive guidelines for the use of synthetic CT in ART [1,4].
This study highlights the potential of synthetic CT derived from CBCT scans to improve the accuracy and efficiency of ART for patients with breast cancer. By reducing additional radiation exposure and enhancing treatment precision, synthetic CT offers a promising solution for optimizing ART in breast cancer treatment. Future research should focus on validating these findings in larger cohorts and expanding the application of synthetic CT to other cancer types and anatomical regions.
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for profit sectors.
The authors have nothing to disclose.
All relevant data are within the paper and its Supporting Information files.
Conceptualization: Sohyun Ahn, Kwangwoo Park. Data curation: So Eun Choi. Formal analysis: So Eun Choi. Investigation: So Eun Choi, Jeong-Heon Kim. Methodology: Sohyun Ahn, Kwangwoo Park. Project administration: Kwangwoo Park, Hai-Jeon Yoon. Resources: Kwangwoo Park. Software: Jeong-Heon Kim. Supervision: Sohyun Ahn, Kwangwoo Park, Hai-Jeon Yoon. Validation: Sohyun Ahn, Kwangwoo Park. Visualization: So Eun Choi. Writing – original draft: Sohyun Ahn, So Eun Choi. Writing – review & editing: Sohyun Ahn, So Eun Choi.
Progress in Medical Physics 2024; 35(4): 145-154
Published online December 31, 2024 https://doi.org/10.14316/pmp.2024.35.4.145
Copyright © Korean Society of Medical Physics.
Sohyun Ahn1,2 , So Eun Choi3
, Jeong-Heon Kim4,5,6
, Kwangwoo Park7
, Hai-Jeon Yoon8
1Ewha Medical Research Institute, Ewha Womans University College of Medicine, Seoul, 2Ewha Medical Artificial Intelligence Research Institute, Ewha Womans University College of Medicine, Seoul, 3Department of Computational Medicine, Ewha Womans University College of Medicine, Seoul, 4Department of Medicine, Yonsei University College of Medicine, Seoul, 5Medical Physics and Biomedical Engineering Lab (MPBEL), Yonsei University College of Medicine, Seoul, 6Department of Radiation Oncology, Yonsei Cancer Center, Heavy Ion Therapy Research Institute, Yonsei University College of Medicine, Seoul, 7Department of Radiation Oncology, Yongin Severance Hospital, Yonsei University College of Medicine, Yongin, 8Department of Nuclear Medicine, Ewha Womans University Mokdong Hospital, Ewha Womans University College of Medicine, Seoul, Korea
Correspondence to:Sohyun Ahn
(mpsohyun@gmail.com)
Tel: 82-2-6986-6305
Fax: 82-0504-158-4052
This is an Open-Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Purpose: This study investigated the use of synthetic computed tomography (CT) images derived from cone beam CT (CBCT) scans to analyze dose changes in breast cancer patients undergoing treatment and to evaluate the optimal timing for implementing adaptive radiotherapy.
Methods: A retrospective analysis was conducted on five breast cancer patients treated with tomotherapy-based volumetric-modulated arc therapy at Yongin Severance Hospital. Each patient received 15 fractions, with doses of 320 centigray (cGy) to the high-dose planning target volume (PTV) and 267 cGy to the low-dose PTV. Planning CT images were acquired using the Aquilion scanner, and CBCT images were captured with the VersaHD linear accelerator’s on-board imager. These images were registered in RayStation using a hybrid deformable image registration method to generate synthetic CT images. Dose distributions were reanalyzed using the synthetic CT images, and dosevolume histogram parameters, including the dose to 95% of the volume (D95) and mean dose (Dmean) for the PTV, as well as D95, Dmean, the percentage of the volume receiving at least 5 Gy (V5) and 10 Gy (V10) for organs-at-risk (OARs), were extracted using MATLAB to assess dose changes during treatment.
Results: For the original plans, the mean D95 for PTV high across all patients was 287.13±31.32 cGy, while for PTV low, it was 245.53±6.21 cGy. In contrast, the adaptive plans yielded a mean D95 of 298.17±12.37 cGy for PTV High and 247.25±4.23 cGy for PTV low. The ART Plan may lead to increased dose exposure in certain structures, such as the spinal cord, while providing targeted improvements in reducing radiation exposure in specific OARs (e.g., contralateral breast and esophagus).
Conclusions: Synthetic CT images generated from CBCT scans provide a fast and efficient means of quantifying dose changes, supporting precise patient care through interfractional evaluation. Future studies will aim to apply this method to other organs and larger patient cohorts.
Keywords: Synthetic computed tomography generation, Adaptive radiotherapy, Breast cancer treatment, Dose-volumetric analysis, Cone beam computed tomography imaging
Breast cancer remains one of the most prevalent malignancies affecting women globally, with high mortality rates despite advancements in early detection and treatment technologies. Radiation therapy plays a pivotal role in breast cancer management, particularly as an adjuvant therapy following surgery to reduce the risk of tumor recurrence and improve survival outcomes [1-3]. Traditional radiation therapy planning is typically based on computed tomography (CT) images obtained during the initial treatment stage. However, anatomical changes or tumor shifts occurring during therapy can compromise the accuracy of the initial treatment plan [3-5].
To mitigate this challenge, adaptive radiation therapy (ART) has been introduced. ART adapts the treatment plan in response to anatomical changes over time, thereby allowing for more precise and effective radiation delivery. Despite its advantages, ART requires multiple CT scans, which can result in additional radiation exposure and increased time and cost burdens for patients.
A promising solution to this problem is the use of synthetic CT derived from cone-beam CT (CBCT). Synthetic CT provides images of similar quality to conventional CT without additional radiation exposure, thereby enabling accurate assessment of anatomical changes in patients. This approach enhances the efficiency of ART and allows for treatment plan optimization [5-7].
The primary aim of this study was to generate synthetic CT images from CBCT images and use them to evaluate treatment accuracy in patients with breast cancer undergoing ART. This approach allows timely adjustments to treatment plans, thereby maximizing therapeutic efficacy while minimizing side effects. Future research will focus on expanding the applicability of this method to other anatomical regions, further demonstrating the robustness and versatility of synthetic CT-based ART evaluation [8,9].
This retrospective study included five female patients with breast cancer who were treated at the Yongin Severance Hospital between March 2020 and July 2023. The patients were treated using tangential volumetric modulated arc therapy (T-VMAT) at doses of 320 centigray (cGy) for planning target volumes (PTV) high and 267 cGy for PTV low, over 15 fractionated sessions. Informed consent was obtained from all participants (Table 1).
Table 1 . Patient characteristics.
Variable | Value |
---|---|
Sex (n) | |
Male | |
Female | 5 |
Age (y) | |
Mean (range) | 53 (40–68) |
Median | 52 |
Fraction, mean | |
Total fractions | 15 |
Number of ARTs (range) | 11 (8–13) |
Planning CT images were acquired using an Aquilion CT scanner (Canon Medical Systems). The patients were placed in the supine position with their arms raised above their heads to simulate the treatment conditions. The scans were performed with a slice thickness of 3 mm to provide the detailed anatomical information necessary for precise treatment planning. CBCT images were obtained using the On-Board Imager system integrated with a VersaHD LINAC (Elekta). These scans were performed before each treatment session to monitor anatomical changes and ensure accurate patient positioning. The CBCT imaging protocol included a full gantry rotation to capture comprehensive volumetric data, which were then used to generate synthetic CT images (Fig. 1).
Synthetic CT images were generated from CBCT scans using a hybrid deformable image registration method within the RayStation treatment planning system (RaySearch Laboratories). This approach involved the following steps:
1) Rigid registration: Initial alignment of CBCT images to planning CT based on bony anatomy.
2) Deformable registration: Fine-tuning the alignment to account for soft tissue variations and patient positioning differences. This process ensures that the critical structures and target volumes are accurately represented.
3) Quality assurance: The generated synthetic CT images were reviewed for accuracy in anatomical representation and were validated against planning CT images.
All treatment plans were developed using RayStation. The following steps were undertaken:
1) Contouring: Delineation of the PTVs and organs-at-risk (OARs) in the planning CT images. The OARs included the breast, heart, ipsilateral lung, spinal cord, esophagus, and contralateral breast.
2) Plan optimization: Optimization of the T-VMAT plans to achieve adequate dose coverage of the PTV while minimizing exposure to OARs. The dose prescriptions were set at 320 cGy for PTV high and 267 cGy for PTV low, delivered over 15 fractions.
3) Verification: Each treatment plan was reviewed and adjusted based on synthetic CT images to account for anatomical changes throughout the treatment course.
Dosimetry parameters were extracted and compared between the original and adaptive treatment plans through synthetic CT-based evaluation using dose–volume histogram (DVH) metrics. The following parameters were analyzed: D95, dose received by 95% of the target volume; Dmean, mean dose received by the target volume; and V5 and V10, volumes of the OARs receiving 5 and 10 Gy, respectively.
The dataset included paired data for each patient, comprising the Original and ART Plans. On average, ART data covered 11 fractions per patient. To robustly analyze the dose differences between the two plans, a linear mixed model (LMM) was applied.
The primary goal of employing LMM was to evaluate the differences between the Original and ART Plans with respect to DVH parameters in the same region of interest.
The LMM used in this study was structured as follows:
(1) Dependent variable: Log-transformed DVH values were selected as the dependent variable to normalize the data distribution and address variability in the ranges.
(2) Independent variable: The treatment plan, which was categorized as either “Original Plan” or “ART Plan,” was modeled as the fixed effect to capture the effects on DVH differences.
(3) Random effect: Patient ID was included as a random effect to adjust for individual patient-specific variability, allowing for generalizable analysis without conflating interpatient differences with treatment effects.
For each patient, the mean dose volume was calculated separately for the ART Plan and Original Plan:
(1) Original Plan mean per patient: The mean of each fraction’s DVH parameters in the Original Plan group.
(2) ART Plan mean per patient: The mean of each fraction’s DVH parameters in the ART Plan group.
(3) Average Original Plan: The sum of the Original Plan Means per Patient divided by the total number of patients.
(4) Average ART Plan: The sum of the ART Plan Means per Patient divided by the total number of patients.
For each patient, the difference between the ART Plan and Original Plan means is calculated as follows:
(1) Difference per patient: The Original Plan Mean per Patient is subtracted from the ART Plan Mean per Patient for each patient.
(2) Mean difference: The sum of the Difference per Patient divided by the total number of patients.
(3) Relative difference (%): The mean difference is divided by the average Original Plan and then multiplied by 100 to express the result as a percentage.
All procedures involving human participants were performed according to the ethical standards of the Institutional and/or National Research Committee and the 2013 Helsinki Declaration and its later amendments or comparable ethical standards.
Based on the synthetic CT-based dosimetry analysis, the D95 and Dmean for PTV high exhibited notable differences between the original and adaptive treatment plans. Table 2 presents the LMM results, and Table 3 presents the
Table 2 . LMM results for PTV and OARs.
ROI | Variable | Group coefficient (Ref. Original Plan) | Group standard error (Ref. Original Plan) | Group variance |
---|---|---|---|---|
PTV high | D95 | 0.043 | 0.010 | 0.006 |
Dmean | 0.013 | 0.003 | 0 | |
PTV low | D95 | 0.007 | 0.005 | 0 |
Dmean | 0.001 | 0.001 | 0 | |
Heart | V5 | −0.006 | 0.010 | 0.005 |
V10 | 0.005 | 0.004 | 0.003 | |
D95 | −0.007 | 0.008 | 0.005 | |
Dmean | 0.009 | 0.009 | 0.003 | |
Spinal cord | V5 | 0.002 | 0.004 | 0.117 |
V10 | 0.031 | 0.006 | 0.028 | |
D95 | 0.175 | 0.040 | 0.086 | |
Dmean | 0.023 | 0.020 | 0.348 | |
Esophagus | V5 | 0.046 | 0.022 | 0.138 |
V10 | −0.010 | 0.002 | 0.016 | |
D95 | 0.017 | 0.048 | 0 | |
Dmean | −0.023 | 0.018 | 0.171 | |
Ipsilateral lung | V5 | −0.004 | 0.006 | 0.047 |
V10 | 0.036 | 0.007 | 0.060 | |
D95 | 0.019 | 0.011 | 0.029 | |
Dmean | 0.017 | 0.009 | 0.110 | |
Contralateral breast | V5 | −0.014 | 0.009 | 0.012 |
V10 | 0.006 | 0.010 | 0.012 | |
D95 | −0.065 | 0.018 | 0.020 | |
Dmean | −0.002 | 0.008 | 0 |
Table 3 . Overall average dose per fraction for PTV and OARs across patients.
ROI | Variable | Original Plan | ART Plan |
---|---|---|---|
PTV high | D95 | 287.13±31.32 | 298.17±12.37 |
Dmean | 316.80±9.50 | 320.88±4.53 | |
PTV low | D95 | 245.53±6.21 | 247.25±4.23 |
Dmean | 274.93±5.69 | 275.20±4.19 | |
Heart | V5 | 0.56±0.14 | 0.55±0.11 |
V10 | 0.06±0.07 | 0.07±0.06 | |
D95 | 6.19±0.42 | 6.16±0.64 | |
Dmean | 17.16±1.55 | 17.27±0.72 | |
Spinal cord | V5 | 0.33±0.53 | 0.33±0.49 |
V10 | 0.12±0.17 | 0.16±0.23 | |
D95 | 0.28±0.33 | 0.67±0.81 | |
Dmean | 8.48±5.51 | 9.04±5.93 | |
Esophagus | V5 | 0.49±0.62 | 0.55±0.54 |
V10 | 0.09±0.15 | 0.07±0.14 | |
D95 | 3.88±1.71 | 3.79±1.04 | |
Dmean | 13.63±5.67 | 13.39±5.68 | |
Ipsilateral lung | V5 | 2.45±0.63 | 2.45±0.72 |
V10 | 1.41±0.57 | 1.49±0.54 | |
D95 | 6.83±1.24 | 7.03±1.50 | |
Dmean | 45.83±14.08 | 46.86±14.77 | |
Contralateral breast | V5 | 0.82±0.26 | 0.79±0.15 |
V10 | 0.16±0.17 | 0.16±0.11 | |
D95 | 6.98±1.47 | 6.49±1.06 | |
Dmean | 18.60±1.28 | 18.53±0.94 |
Patient-specific dose distribution differences were also analyzed. Patient 1, who exhibited the smallest deviation between the original and adaptive treatment plans, exhibited relatively stable D95 values across both PTV high and PTV low. In contrast, Patient 4 had the largest difference between the two treatment plans, with significant variation in D95 values, reflecting the need for patient-specific consideration when performing ART. Fig. 3 presents the comparison of D95 and Dmean values for high PTV in Patients 1 and 4, emphasizing the role of synthetic CT-based plan evaluation in accommodating patient-specific differences.
The analysis of PTV coverage revealed significant improvements in dose conformity with the ART Plan compared with the Original Plan (Table 4). In particular,
Table 4 . Relative differences between the Original Plan and ART Plan for PTV and OARs.
ROI | Variable | Mean difference | Relative difference (%) | |
---|---|---|---|---|
PTV high | D95 | 11.04±19.38 | 3.85 | <0.001 |
Dmean | 4.08±8.45 | 1.29 | <0.001 | |
PTV low | D95 | 1.72±4.95 | 0.70 | 0.217 |
Dmean | 0.27±1.72 | 0.10 | 0.169 | |
Heart | V5 | −0.01±0.14 | −2.32 | 0.525 |
V10 | 0.01±0.03 | 7.82 | 0.142 | |
D95 | −0.03±0.47 | −0.47 | 0.370 | |
Dmean | 0.12±1.30 | 0.68 | 0.316 | |
Spinal cord | V5 | −0.01±0.05 | −1.74 | 0.614 |
V10 | 0.04±0.07 | 35.72 | <0.001 | |
D95 | 0.39±0.70 | 136.00 | <0.001 | |
Dmean | 0.55±1.41 | 6.50 | 0.271 | |
Esophagus | V5 | 0.06±0.30 | 12.04 | 0.042 |
V10 | −0.01±0.02 | −14.56 | <0.001 | |
D95 | −0.09±1.19 | −2.29 | 0.729 | |
Dmean | −0.24±2.40 | −1.78 | 0.195 | |
Ipsilateral lung | V5 | 0.00±0.20 | 0.18 | 0.535 |
V10 | 0.08±0.13 | 5.73 | <0.001 | |
D95 | 0.19±0.75 | 2.80 | 0.082 | |
Dmean | 1.03±4.06 | 2.25 | 0.067 | |
Contralateral breast | V5 | −0.04±0.16 | −4.43 | 0.102 |
V10 | 0.00±0.13 | 1.25 | 0.567 | |
D95 | −0.49±1.19 | −7.02 | <0.001 | |
Dmean | −0.07±1.66 | −0.37 | 0.813 |
(1) D95: The ART Plan demonstrated a significantly higher dose difference in PTV high (average 3.85%,
(2) Dmean: For PTV high, the ART Plan achieved a marginal reduction in Dmean, with a statistically significant difference (
These findings suggest that the ART Plan significantly enhances dose conformity in PTV high, thereby ensuring robust target coverage while maintaining precision in dose delivery.
We conducted a dosimetry analysis of the OAR structures, particularly the heart and ipsilateral lung, using synthetic CT-based evaluation across the cohort (Fig. 4 and Table 4).
(1) V5: ART exhibited minor improvements in reducing dose exposure for most OARs; however, these changes were not statistically significant. In contrast, the esophagus exhibited a significant dose increase (12.04%); this change was statistically significant (
(2) V10: ART achieved significant reductions in dose exposure for the esophagus (−14.56%). In contrast, significant dose increases were observed in the spinal cord (35.72%) and ipsilateral lung (5.73%) (
(3) D95: ART exhibited a significant reduction in dose exposure for the contralateral breast (average −7.02%,
(4) Dmean: Minor dose increases were noted for the spinal cord and ipsilateral lung, whereas the other OARs exhibited minor dose decreases. None of the average differences were statistically significant.
These findings suggest that the ART Plan increases dose exposure in certain structures, such as the spinal cord, while providing targeted improvements in reducing radiation exposure in specific OARs (e.g., contralateral breast and esophagus).
This study demonstrated the efficacy of using synthetic CT images derived from CBCT scans to guide ART for patients with breast cancer. The relative differences in DVH parameters, such as the increase in D95 for the high-dose PTV region from 287.13±31.32 cGy in the Original Plan to 298.17±12.37 cGy in the adaptive plan, highlight the importance of synthetic CT-based plan evaluation in enhancing treatment accuracy and dose conformity.
Moreover, the trends observed in the OARs underscore the complex effects of adaptive planning. Although the ART Plan effectively reduced dose exposure in specific structures, it also increased exposure in other structures, particularly the spinal cord. This highlights the necessity of precise adaptive planning, supported by inter-fractional evaluation, to mitigate potential risks. These findings suggest that ART can optimize the balance between target coverage and organ sparing but requires careful monitoring of dose tradeoffs to critical structures.
Synthetic CT in ART offers several clinical advantages. First, it reduces the need for repeated CT scans, thereby minimizing additional radiation exposure and associated risks for patients. This approach also enhances workflow efficiency by providing rapid and reliable anatomical assessments, allowing for timely treatment plan adjustments [2,10].
The significant dosimetry differences observed suggest that synthetic CT-based evaluation can enhance the precision of radiation delivery, potentially enhancing tumor control and reducing side effects. The observed differences in DVH parameters between the original and adaptive treatment plans indicate that synthetic CT can effectively capture dose distribution changes, thereby ensuring consistent and accurate treatment delivery [1,4,11].
Despite these promising findings, this study has several limitations. The sample size of five patients was relatively small, which may limit the generalizability of the results. Furthermore, the study was retrospective, which could introduce selection bias. Future studies should include larger, prospective cohorts to validate these findings and further investigate the benefits of synthetic CT-based plan evaluation in ART [3,12].
Another limitation is the potential for variability in synthetic CT generation due to differences in CBCT image quality and patient positioning. Standardizing the imaging and registration protocols could help reduce this variability and ensure more consistent results across different clinical settings [6,13].
Future research should focus on expanding the application of synthetic CT in ART to other anatomical regions and cancer types. Investigating the long-term clinical outcomes of patients treated with adaptive plan evaluation based on synthetic CT will provide valuable insights into the efficacy and safety of this approach. Furthermore, integrating advanced imaging techniques and machine learning algorithms can enhance the accuracy and efficiency of synthetic CT generation and dose prediction [2,10,11].
The development of robust, standardized protocols for synthetic CT generation and validation is crucial for widespread clinical adoption. Collaborative efforts between research institutions and clinical centers can facilitate the sharing of best practices and contribute to the development of comprehensive guidelines for the use of synthetic CT in ART [1,4].
This study highlights the potential of synthetic CT derived from CBCT scans to improve the accuracy and efficiency of ART for patients with breast cancer. By reducing additional radiation exposure and enhancing treatment precision, synthetic CT offers a promising solution for optimizing ART in breast cancer treatment. Future research should focus on validating these findings in larger cohorts and expanding the application of synthetic CT to other cancer types and anatomical regions.
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for profit sectors.
The authors have nothing to disclose.
All relevant data are within the paper and its Supporting Information files.
Conceptualization: Sohyun Ahn, Kwangwoo Park. Data curation: So Eun Choi. Formal analysis: So Eun Choi. Investigation: So Eun Choi, Jeong-Heon Kim. Methodology: Sohyun Ahn, Kwangwoo Park. Project administration: Kwangwoo Park, Hai-Jeon Yoon. Resources: Kwangwoo Park. Software: Jeong-Heon Kim. Supervision: Sohyun Ahn, Kwangwoo Park, Hai-Jeon Yoon. Validation: Sohyun Ahn, Kwangwoo Park. Visualization: So Eun Choi. Writing – original draft: Sohyun Ahn, So Eun Choi. Writing – review & editing: Sohyun Ahn, So Eun Choi.
Table 1 Patient characteristics
Variable | Value |
---|---|
Sex (n) | |
Male | |
Female | 5 |
Age (y) | |
Mean (range) | 53 (40–68) |
Median | 52 |
Fraction, mean | |
Total fractions | 15 |
Number of ARTs (range) | 11 (8–13) |
Table 2 LMM results for PTV and OARs
ROI | Variable | Group coefficient (Ref. Original Plan) | Group standard error (Ref. Original Plan) | Group variance |
---|---|---|---|---|
PTV high | D95 | 0.043 | 0.010 | 0.006 |
Dmean | 0.013 | 0.003 | 0 | |
PTV low | D95 | 0.007 | 0.005 | 0 |
Dmean | 0.001 | 0.001 | 0 | |
Heart | V5 | −0.006 | 0.010 | 0.005 |
V10 | 0.005 | 0.004 | 0.003 | |
D95 | −0.007 | 0.008 | 0.005 | |
Dmean | 0.009 | 0.009 | 0.003 | |
Spinal cord | V5 | 0.002 | 0.004 | 0.117 |
V10 | 0.031 | 0.006 | 0.028 | |
D95 | 0.175 | 0.040 | 0.086 | |
Dmean | 0.023 | 0.020 | 0.348 | |
Esophagus | V5 | 0.046 | 0.022 | 0.138 |
V10 | −0.010 | 0.002 | 0.016 | |
D95 | 0.017 | 0.048 | 0 | |
Dmean | −0.023 | 0.018 | 0.171 | |
Ipsilateral lung | V5 | −0.004 | 0.006 | 0.047 |
V10 | 0.036 | 0.007 | 0.060 | |
D95 | 0.019 | 0.011 | 0.029 | |
Dmean | 0.017 | 0.009 | 0.110 | |
Contralateral breast | V5 | −0.014 | 0.009 | 0.012 |
V10 | 0.006 | 0.010 | 0.012 | |
D95 | −0.065 | 0.018 | 0.020 | |
Dmean | −0.002 | 0.008 | 0 |
Table 3 Overall average dose per fraction for PTV and OARs across patients
ROI | Variable | Original Plan | ART Plan |
---|---|---|---|
PTV high | D95 | 287.13±31.32 | 298.17±12.37 |
Dmean | 316.80±9.50 | 320.88±4.53 | |
PTV low | D95 | 245.53±6.21 | 247.25±4.23 |
Dmean | 274.93±5.69 | 275.20±4.19 | |
Heart | V5 | 0.56±0.14 | 0.55±0.11 |
V10 | 0.06±0.07 | 0.07±0.06 | |
D95 | 6.19±0.42 | 6.16±0.64 | |
Dmean | 17.16±1.55 | 17.27±0.72 | |
Spinal cord | V5 | 0.33±0.53 | 0.33±0.49 |
V10 | 0.12±0.17 | 0.16±0.23 | |
D95 | 0.28±0.33 | 0.67±0.81 | |
Dmean | 8.48±5.51 | 9.04±5.93 | |
Esophagus | V5 | 0.49±0.62 | 0.55±0.54 |
V10 | 0.09±0.15 | 0.07±0.14 | |
D95 | 3.88±1.71 | 3.79±1.04 | |
Dmean | 13.63±5.67 | 13.39±5.68 | |
Ipsilateral lung | V5 | 2.45±0.63 | 2.45±0.72 |
V10 | 1.41±0.57 | 1.49±0.54 | |
D95 | 6.83±1.24 | 7.03±1.50 | |
Dmean | 45.83±14.08 | 46.86±14.77 | |
Contralateral breast | V5 | 0.82±0.26 | 0.79±0.15 |
V10 | 0.16±0.17 | 0.16±0.11 | |
D95 | 6.98±1.47 | 6.49±1.06 | |
Dmean | 18.60±1.28 | 18.53±0.94 |
Table 4 Relative differences between the Original Plan and ART Plan for PTV and OARs
ROI | Variable | Mean difference | Relative difference (%) | |
---|---|---|---|---|
PTV high | D95 | 11.04±19.38 | 3.85 | <0.001 |
Dmean | 4.08±8.45 | 1.29 | <0.001 | |
PTV low | D95 | 1.72±4.95 | 0.70 | 0.217 |
Dmean | 0.27±1.72 | 0.10 | 0.169 | |
Heart | V5 | −0.01±0.14 | −2.32 | 0.525 |
V10 | 0.01±0.03 | 7.82 | 0.142 | |
D95 | −0.03±0.47 | −0.47 | 0.370 | |
Dmean | 0.12±1.30 | 0.68 | 0.316 | |
Spinal cord | V5 | −0.01±0.05 | −1.74 | 0.614 |
V10 | 0.04±0.07 | 35.72 | <0.001 | |
D95 | 0.39±0.70 | 136.00 | <0.001 | |
Dmean | 0.55±1.41 | 6.50 | 0.271 | |
Esophagus | V5 | 0.06±0.30 | 12.04 | 0.042 |
V10 | −0.01±0.02 | −14.56 | <0.001 | |
D95 | −0.09±1.19 | −2.29 | 0.729 | |
Dmean | −0.24±2.40 | −1.78 | 0.195 | |
Ipsilateral lung | V5 | 0.00±0.20 | 0.18 | 0.535 |
V10 | 0.08±0.13 | 5.73 | <0.001 | |
D95 | 0.19±0.75 | 2.80 | 0.082 | |
Dmean | 1.03±4.06 | 2.25 | 0.067 | |
Contralateral breast | V5 | −0.04±0.16 | −4.43 | 0.102 |
V10 | 0.00±0.13 | 1.25 | 0.567 | |
D95 | −0.49±1.19 | −7.02 | <0.001 | |
Dmean | −0.07±1.66 | −0.37 | 0.813 |
pISSN 2508-4445
eISSN 2508-4453
Formerly ISSN 1226-5829
Frequency: Quarterly