Main Session
Sep
29
PQA 07 - Head and Neck Cancer, Lung Cancer/Thoracic Malignancies, and Nursing and Supportive Care
Presenter(s)
Min Liu, PhD - Sichuan Cancer Hospital, Chengdu, Sichuan
M. Liu, X. Liao Sr, B. Tang, Y. Wang, W. Xianliang, and L. C. Orlandini; Sichuan Cancer Hospital and Research Institute, University of Electronic Science and Technology of China, Chengdu, China
Purpose/Objective(s):
In the 1.5T Unity MR-guided adaptive workflow, synthetic CT (sCT) generation uses structure-wise bulk density assignment, assigning a single mean electron density (ED) per organ. sCT recalculation in lung SBRT often fails to reproduce CT-based dose distributions, indicating a limitation of lung density averaging. We hypothesized that unmodeled high-density intrapulmonary vascular substructures are a primary physical source of these discrepancies.Materials/Methods:
Fifty-one lung SBRT cases (50 Gy in 5 fractions) demonstrating DVH deviations between the CT reference plan (TPref) and sCT-based recalculation under conventional bulk density assignment (TPsCT) were retrospectively analyzed. Three categories of vascular-related intrapulmonary substructures (blood vessels, pulmonary artery, and pulmonary venous system) were automatically segmented and defined as separate regions of interest. Patient-specific relative electron density (EDr) values were extracted from the simulation CT and assigned to these structures, thereby excluding them from the homogeneous lung parenchyma bulk assignment. A third treatment plan (TPsCT_sub) was recalculated on the sCT incorporating substructure-specific EDr overrides. All three plans were compared using target DVH dosimetric parameters.Results:
Vascular substructures accounted for a median 8.3% of total lung volume (5.2–13.6%) and exhibited substantially higher EDr than lung parenchyma (0.52, 1.05, and 1.10 for vessels, pulmonary artery, and pulmonary venous system, respectively). Excluding these components reduced median lung parenchymal EDr from 0.27 to 0.23 confirming a physical basis for bulk-assignment error. Target-dose agreement relative to the TPref is summarized in Table 1. Under TPsCT, only 25.5% of patients achieved =2% agreement for V50Gy, compared with 78.4% using TPsCT_sub. For D50%, agreement improved from 27.5% to 80.4%. Similar trends were observed for D20%. TPsCT_sub significantly reduced absolute target-dose deviation (Wilcoxon signed-rank, p < 0.001).Conclusion:
Substructure-guided, patient-specific EDr assignment reduced sCT dose deviations and improved agreement with CT-based dosimetry in most lung SBRT cases. This approach offers a feasible and physically grounded strategy for refining density modeling within the standard MR-guided adaptive workflow.| Parameter | TPref | TPsCT | TPsCT_sub | Patients within Ddiff < 2% (%) Bulk/Sub |
| V50Gy (%) | 95.1 (72.0-96.0) | 97.1 (75.3-99.9) | 96.1 (72.2-99.0) | 25.5/78.4 |
| D50% (Gy) | 58.4 (53.4-61.5) | 59.7 (54.1-64.6) | 58.5 (53.6-61.6) | 27.5/80.4 |
| D20% (Gy) | 62.9 (56.1-65.6) | 63.8 (57.6-68.4) | 63.0 (56.5-66.1) | 49.0/80.4 |