BLUE RIBBON POSTER IMAGING: Investigating Low Dose CT for Ventilation Imaging
Abstract
Purpose
Functional imaging of the lung has found multiple clinical applications such as improving detection of chronic obstructive pulmonary disease and improving outcomes in radiation therapy by sparing healthy tissue. CT Ventilation Imaging (CTVI) is a recently approved lung function imaging modality with high accessibility, however work to date has focused on standard-dose CT. This study evaluates the impact of dose reduction on the accuracy of CTVI.
Methods
Paired inhale-exhale breath-hold CT images from two lung cancer cohorts (N = 28) were used as inputs to a validated noise-addition software to generate reduced dose images at 50%, 25%, 10%, and 5% relative dose. Ventilation images were derived from deformable image registration evaluation of volume change (CTVIJac) or change in Hounsfield Unit (HU) value (CTVIHU). Low-dose CT ventilation images were compared against standard-dose using voxel-wise Pearson correlation, and Dice similarity coefficient (DSC) of regions of interest defined by upper 85% (functional lung) and upper 15% (highly functioning lung) of ventilation values.
Results
CTVIJac consistently outperformed CTVIHU, retaining very strong correlations with standard-dose images at all relative dose levels. As relative dose decreased from 50% to 5%, mean correlations decreased from 0.97 to 0.85 (CTVIJac), and 0.96 to 0.76 (CTVIHU); mean DSC of functional lung decreased from 0.98 to 0.94 (CTVIJac) and 0.95 to 0.90 (CTVIHU); mean DSC of highly functioning lung decreased from 0.90 to 0.68 (CTVIJac), 0.86 to 0.63 (CTVIHU). Dose-dependent trends were consistent across scanners and patient cohorts.
Conclusion
CTVI is robust to dose reduction, particularly when using volume-based ventilation metrics (CTVIJac). These results support the investigation of CTVI for screening and longitudinal studies, and the potential for significant dose optimization in CTVI workflows.