Click here to

Session: New and Emerging Technology [Return to Session]

Characterization of Functional Lung Tissue

M Rodriguez1,2, A Wright1,2, WT Watkins1*, (1) City of Hope Medical Center, Duarte, CA (2) University of California Riverside, Riverside, CA

Presentations

SU-H330-IePD-F8-4 (Sunday, 7/10/2022) 3:30 PM - 4:00 PM [Eastern Time (GMT-4)]

Exhibit Hall | Forum 8

Purpose: To evaluate and correlate lung function to lung density and lung deformation.

Methods: A dataset of 50-paired 4DCT and functional lung image sets were evaluated (the VAMPIRE dataset) including 46-human images and 4-sheep. This dataset was a validation challenge between functional imaging and 4DCT. We evaluated per-patient correlations with functional imaging, CT-density, and deformable image registration (DIR) between exhale- and inhale- images. Elastix image registration software was scripted using rigid (inter-modality) and deformable B-spline registration in 4DCT; Jacobian and CT-numbers were extracted from the DVF on the exhale scan, and ventilation calculations included both DVF and voxel-density information. Functional lung images were normalized into 10-bins representing low-to-high functionality from images including Xenon-CT, 99mTc-DTPA-SPECT, and Galligas-PET on order to consistently scale the various modalities. We evaluated patient- and population- correlations between tissue functionality, density, and deformation.

Results: The volume of functional lung labeled with at least 25% functionality was consistent in DTPASPECT (3400cc±1400cc) and Ga-PET (3600cc±900cc). Correlation between lung function and density, JacDet, and Ventilation were consistent with the VAMPIRE study (~0.5). There was not a statistically significant difference between ventilation and JacDet; mean and standard deviation of the correlation coefficient between the 2 predictors was 0.44+0.11 and 0.45+0.11. Ventilation scales the volume change according to density change, however lung density in a perfect 1-to-1 mapping of deformable lung tissue mass and CT-density should be conserved. A stepwise linear regression model found both CT-number and Jacobian significant (p<10-6) in predicting functionality, with a 2-parameter model: -0.0016*CT+26*JacDet, however, significant inter-patient variations were observed. In patients with relatively high correlation in JacDet and functionality (r2>0.6 in 28 patients), CT-numbers in the range of -850 to -500 and 5-10% deformation appear predictive.

Conclusion: Lung function derived from CT remains elusive. We observed significant inter-patient variations in relationships between functional tissue, CT-number, and deformation.

Keywords

Functional Imaging, Lung

Taxonomy

IM/TH- Image Analysis (Single Modality or Multi-Modality): Image processing

Contact Email

Share: