Room: Track 2
Purpose: To test the efficacy of 5DCT model generation and biomechanical property estimation with low-dose images for diagnostic applications.
Methods: The 5DCT protocol utilizes a combination of 25 fast-helical free breathing CT (FHFBCT) and simultaneous breathing monitoring by an abdominal bellows. One scan is defined as the reference scan, and the other 24 scans are deformably mapped to it. The voxel positions are coupled with the breathing amplitudes and rates to fit a voxel-specific motion modeling equation. Two CT images are generated using the 5D model at end-exhalation and end-inhalation. The deformable vector field from end-exhalation to end-inhalation is used to estimate tissue elasticity from a well-validated finite element biomechanical lung model, yielding the lung tissue elasticity distribution. The total dose from the FHFBCT protocol is currently 30 mSv from 25 scans using 40 mAs. In this study, we injected noise to simulate 2 mAs scans for a total 1.5 mSv. This dose is well below typical diagnostic CT doses, and thus if model discrepancies are low, biomechanical property estimation application research could be justified.
Results: The 5DCT-generated deformation vectors for the nominal and noise-injected datasets were evaluated by vector magnitude and angle differences, which were on average 0.820 mm and 8.25 degrees, respectively. The difference between the nominal and noise-injected average volume elasticities were 0.760 kPa (typical parenchymal tissue is 1-8 kPa). Thus, the low-dose FHFBCT protocol could be used to provide deformation and subsequent elasticity estimations.
Conclusion: We have measured the effect of reducing imaging dose to the patient in our biomechanical property estimation workflow. This work opens the door to the testing of low dose and ultra-low dose diagnostic free breathing CT scans for 5DCT reconstruction and elastography.