Premium
SU‐D‐BRB‐01: 4D‐CT Lung Ventilation Images Vary with 4D‐CT Sorting Techniques
Author(s) -
Yamamoto T,
Kabus S,
Lorenz C,
Johnston E,
Maxim P,
Loo B,
Keall P
Publication year - 2012
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4734673
Subject(s) - image registration , voxel , hounsfield scale , nuclear medicine , ventilation (architecture) , medical imaging , medicine , sorting , radiology , artificial intelligence , computed tomography , computer science , image (mathematics) , algorithm , physics , thermodynamics
Purpose: 4D‐CT ventilation imaging is a novel promising technique for lung functional imaging and has potential as a biomarker for radiation pneumonitis, but has not been validated in human subjects. The current 4D‐ CT technique with phase‐based sorting results in artifacts at an alarmingly high frequency (90%), which may introduce variations into ventilation calculations. The purpose of this study was to quantify the variability of 4D‐ CT ventilation imaging to 4D‐CT sorting techniques. Methods: Two 4D‐CT images were generated from the same data set by: (1) phase‐based; (2) anatomic similarity‐ and abdominal displacement‐based sorting for five patients. Two ventilation image sets (V_phase and V_anat) were then calculated by deformable image registration of peak‐exhale and peak‐inhale4D‐CT images and quantification of regional volume change based on Hounsfield unit change. The variability of 4D‐CT ventilation imaging wasquantified using the voxel‐based Spearman rank correlation coefficients and Dice similarity coefficients (DSC) for the spatial overlap of segmented low‐ functional lung regions. The relationship between the abdominal motionrange variation and ventilation variation was also assessed using linearregression. Furthermore, the correlations between V_phase or V_anat and SPECT ventilation images (assumed ground‐truth) were compared. Results: In general, displacement‐ and anatomic similarity‐based sorting reduced 4D‐ CT artifacts compared to phase‐based sorting. The voxel‐based correlationsbetween V_phase and V_anat were only moderate (range, 0.57–0.77). The DSCs for the low‐functional lung regions were moderate to substantial (0.58–0.70). The relationship between the motion range variation and ventilation variation was strong on average (R2=0.79±0.25), suggesting that ventilation variations are related to 4D‐CT artifacts. Vanat was found to improve correlations with SPECT ventilation images compared to V_phase. Conclusions: 4D‐CT ventilation images vary markedly with 4D‐CT sorting techniques. 4D‐CT artifacts should be considered as a significant source of variation in 4D‐CT ventilation imaging during its validation. This study wassupported in part by NIH/NCI R01 93626. SK and CL are employees ofPhilips Research.