2181 - 4D-CT Ventilation and SPECT Perfusion Image-guided Radiotherapy with Knowledge-Based Planning

Purpose/Objective(s): Lung function-guided radiotherapy has the potential to reduce the risk of pulmonary toxicity. However, it is associated with challenges such as functional variations and limited avoidance experience. Knowledge-based planning (KBP) enhances uniformity and efficiency by learning from treatment plans. Despite its advantages, its application in ventilation and perfusion imaging remains underexplored. Herein, we evaluated the quality of KBP plans by incorporating both ventilation and perfusion images.

Materials/Methods: Pretreatment 4D-CT and ^99mTc-macroaggregated albumin SPECT perfusion images from 46 patients with non-small cell lung cancer who received radiotherapy were analyzed. Ventilation images were generated by evaluating changes in CT values through deformable image registration between inspiratory and expiratory images. Regions with the top 33% of function in both ventilation and perfusion images were defined as highly functional regions. The manual treatment plan (MP) was created to avoid these highly functional regions. The MP was designed to meet the dose constraints of RTOG0617 (60 Gy in 30 fractions, D_95% prescription). The data were divided into 33 training and 13 testing cases. A threefold cross-validation method was used to optimize the KBP model using the training dataset. The KBP model was then applied to the testing dataset to generate KBP plans in a single optimization process. Evaluation indices included mean lung dose (MLD), V₂₀ in highly ventilated or highly perfused regions, mean dose (D_mean) of the esophagus, and maximum dose (D_max) of the spinal cord. Statistical differences between MP and KBP were determined using the t-test or Wilcoxon signed-rank test (significance level = 0.05).

Results: All plans met the RTOG0617 dose constraints. For highly ventilated regions, the MLD and V₂₀ for MP and KBP were 12.2 Gy vs 13.5 Gy (P = 0.13) and 20.2% vs 23.3% (P = 0.33), respectively. For highly perfused regions, the MLD and V₂₀ for MP and KBP were 9.9 Gy vs. 11.1 Gy (P = 0.18) and 14.1% vs. 15.9% (P = 0.94), respectively. The conformity index and homogeneity index for the planning target volume for MP and KBP were 0.81 vs 0.78 (P = 0.014) and 1.09 vs 1.06 (P < 0.01), respectively. D_mean of the esophagus and D_max of the spinal cord were 22.2 Gy vs 21.2 Gy (P = 0.89) and 43.3 Gy vs 39.7 Gy (P = 0.51), respectively.

Conclusion: In lung cancer radiotherapy, when avoiding highly ventilated and highly perfused regions, KBP achieved dose indices almost equivalent to those of MP while meeting the RTOG0617 constraints, suggesting that KBP has potential clinical applications in lung cancer radiotherapy.