The Comprehensive Cancer Centre of Nanjing Drum Tower Hospital, The Affiliated Hospital of Nanjing University Medical School, Nanjing, 210000, Jiangsu, China.
School of Computer Science and Engineering, Southeast University, Nanjing, 210000, Jiangsu, China.
Jpn J Radiol. 2024 Jul;42(7):765-776. doi: 10.1007/s11604-024-01550-2. Epub 2024 Mar 27.
Radiotherapy planning incorporating functional lung images has the potential to reduce pulmonary toxicity. Free-breathing 4DCT-derived ventilation image (CTVI) may help quantify lung function. This study introduces a novel deep-learning model directly translating planning CT images into CTVI. We investigated the accuracy of generated images and the impact on functional avoidance planning.
Paired planning CT and 4DCT scans from 48 patients with NSCLC were randomized to training (n = 41) and testing (n = 7) data sets. The ventilation maps were generated from 4DCT using a Jacobian-based algorithm to provide ground truth labels (CTVI). A 3D U-Net-based model was trained to map CT to synthetic CTVI (CTVI) and validated using fivefold cross-validation. The highest-performing model was applied to the testing set. Spearman's correlation (r) and Dice similarity coefficient (DSC) determined voxel-wise and functional-wise concordance between CTVI and CTVI. Three plans were designed per patient in the testing set: one clinical plan without CTVI and two functional avoidance plans combined with CTVI or CTVI, aimed at sparing high-functional lungs defined as the top 50% of the percentile ventilation ranges. Dose-volume (DVH) parameters regarding the planning target volume (PTV) and organs at risk (OARs) were recorded. Radiation pneumonitis (RP) risk was estimated using a dose-function (DFH)-based normal tissue complication probability (NTCP) model.
CTVI showed a mean rs value of 0.65 ± 0.04 compared to CTVI. Mean DSC values over the top 50% and 60% of ventilation ranges were 0.41 ± 0.07 and 0.52 ± 0.10, respectively. In the test set (n = 7), all patients' RP-risk benefited from CTVI-guided plans (Risk: 29.24% vs. 49.12%, P = 0.016), and six patients benefited from CTVI-guided plans (Risk: 31.13% vs. 49.12%, P = 0.022). There were no significant differences in DVH and DFH metrics between CTVI and CTVI-guided plan (P > 0.05).
Using deep-learning techniques, CTVI generated from planning CT exhibited a moderate-to-high correlation with CTVI. The CTVI-guided plans were comparable to the CTVI-guided plans, effectively reducing pulmonary toxicity in patients while maintaining acceptable plan quality. Further prospective trials are needed to validate these findings.
纳入功能肺部图像的放射治疗计划有可能降低肺部毒性。自由呼吸 4DCT 衍生的通气图像（CTVI）可能有助于量化肺功能。本研究介绍了一种直接将计划 CT 图像转换为 CTVI 的新型深度学习模型。我们研究了生成图像的准确性以及对功能回避计划的影响。
48 例 NSCLC 患者的配对计划 CT 和 4DCT 扫描被随机分为训练（n=41）和测试（n=7）数据集。使用基于雅可比的算法从 4DCT 生成通气图，以提供地面真实标签（CTVI）。使用基于 3D U-Net 的模型对 CT 进行映射以生成合成 CTVI（CTVI），并使用五重交叉验证进行验证。将表现最佳的模型应用于测试集。Spearman 相关系数（r）和 Dice 相似系数（DSC）确定了 CTVI 和 CTVI 之间的体素和功能一致性。在测试集中，为每位患者设计了三个计划：一个不使用 CTVI 的临床计划和两个结合 CTVI 或 CTVI 的功能回避计划，旨在保护定义为通气范围的前 50%的高功能肺部。记录了与计划靶区（PTV）和危及器官（OAR）相关的剂量体积（DVH）参数。使用基于剂量-功能（DFH）的正常组织并发症概率（NTCP）模型估计放射性肺炎（RP）风险。
CTVI 与 CTVI 相比，平均 rs 值为 0.65±0.04。在通气范围的前 50%和 60%上，平均 DSC 值分别为 0.41±0.07 和 0.52±0.10。在测试集中（n=7），所有患者的 RP 风险均受益于 CTVI 指导的计划（风险：29.24%比 49.12%，P=0.016），6 名患者受益于 CTVI 指导的计划（风险：31.13%比 49.12%，P=0.022）。CTVI 和 CTVI 指导的计划之间的 DVH 和 DFH 指标没有显著差异（P>0.05）。
使用深度学习技术，从计划 CT 生成的 CTVI 与 CTVI 显示出中等至高的相关性。CTVI 指导的计划与 CTVI 指导的计划相当，在保持可接受的计划质量的同时，有效降低了患者的肺部毒性。需要进一步的前瞻性试验来验证这些发现。
