Author: Tatsuya Fujisaki, Hiraku Fuse, Shin Miyakawa, Hiroki Nosaka, Masato Takahashi, Kenji Yasue 👨🔬
Affiliation: Ibaraki Prefectural University of Health Sciences 🌍
Purpose: This study aimed to evaluate the robustness of Jacobian-based and HUchange-based CT ventilation imaging (CTVI) against respiratory amplitude variations using a custom-designed alveolar phantom with a respiratory amplitude control function.
Methods: A non-rigid alveolar phantom with a respiratory amplitude control function was used. The respiratory cycle of the phantom was set to 3.5 seconds, and 4DCT imaging was performed to obtain sets of CT images with two different respiratory amplitudes (CTSmall and CTLarge). For each respiratory amplitude, deformable image registration using the publicly available software Elastix was applied to deform the inspiratory CT to match the expiratory CT. From these deformation fields, Jacobian-based CTVI Small and CTVILarge were created. HU change-based CTVI Small and CTVILarge were created from the reference CT image and the deformed CT image. Each CTVI was categorized into three functional levels—high, middle, and low ventilation—using an equal (33%) distribution. To evaluate the robustness of functional regions between respiratory amplitudes, regions classified as high ventilation in CTVI Small were examined to determine whether they were also classified as high ventilation in CTVILarge. Statistical analysis was performed using Cohen’s Kappa coefficient to assess agreement between the two amplitudes.
Results: For Jacobian-based CTVI, 20.84% and 19.51% of regions were consistently classified as high and low ventilation, respectively, between CTVILarge and CTVISmall. For HU change-based CTVI, these proportions were 17.46% and 13.46%. Cohen's Kappa values were 0.31 for Jacobian-based and 0.19 for HU change-based CTVI (p < 0.001), indicating higher robustness of Jacobian-based CTVI to respiratory amplitude variations.
Conclusion: By improving a custom-designed alveolar phantom with a respiratory amplitude control function, this study assessed the robustness of CTVI against respiratory amplitude variations. The findings suggest that Jacobian-based CTVI is more robust than HU change-based CTVI, providing more consistent and reliable ventilation imaging despite amplitude differences during lung cancer radiotherapy.