Four-dimensional computed tomography-based ventilation imaging in intensity-modulated radiation therapy treatment planning for pulmonary functional avoidance

Abstract

Purpose: To incorporate four-dimensional computed tomography (4DCT)-based ventilation imaging into intensity-modulated radiation therapy (IMRT) treatment planning for pulmonary functional avoidance.

Methods and materials: Nineteen locally advanced lung cancer patients are retrospectively studied. 4DCT images are employed to create ventilation maps for each patient via a density-change-based algorithm with mass correction. The regional ventilation is directly incorporated into the mathematical formulation of a direct aperture optimization model in IMRT treatment planning to achieve functional avoidance and a voxel-based treatment plan. The proposed functional avoidance planning and voxel-based planning are compared to the conventional treatment planning approach purely based on the anatomy of patients. Paired sample t-tests are conducted to see whether dosimetric differences among the three approaches are significant.

Results: Similar planning target volume (PTV) coverage is achieved by anatomical, functional avoidance, and voxel-based approaches. The voxel-based treatment planning performs better than both functional avoidance and anatomical planning to the lung. For a total lung, the average volume reductions in a functional avoidance plan from an anatomical plan, a voxel-based plan from an anatomical plan, and a voxel-based plan from a functional avoidance plan are 7.0%, 16.8%, and 10.6%, respectively for V₄₀ ; and 0.4%, 6.4%, and 6.0%, respectively for mean Lung Dose (MLD). For a functional lung, the reductions are 8.8%, 17.2%, and 9.2%, respectively, for fV₄₀ ; and 1.1%, 6.2%, and 5.2%, respectively, for functional mean lung dose (fMLD). These reductions are obtained without significantly increasing doses to other organs-at-risk. All the pairwise treatment planning comparisons for both total lung and functional lung are statistically significant (p-value $<\alpha =0.05$ ) except for the functional avoidance plan with the anatomical plan pair in which the p-value $>\alpha =0.05$ . From these results, we can conclude that voxel-based treatment planning outperforms both anatomical and functional-avoidance planning.

Conclusions: We propose a treatment planning framework that directly utilizes functional images and compares voxel-based treatment planning with functional avoidance and anatomical treatment planning.

Keywords: column generation algorithm; lung cancer optimization; voxel based treatment.