Evaluate Deformable Image Registration for the Lungs Using Hyperpolarized Gas Tagging MRI
Y Liu1*, W Miller2, F Yin3, J Cai4, (1) Duke University, Durham, North Carolina, (2) University of Virginia, Charlottesville, VA, (3) Duke University Medical Center, Durham, NC, (4) Duke University Medical Center, Durham, NCTU-C-141-8 Tuesday 10:30AM - 12:30PM Room: 141
Purpose: To evaluate deformable image registration for the lungs using hyperpolarized gas MR tagging technique.
Method and Materials: A 3D MR grid-tagging technique using hyperpolarized helium-3 (HP-He3) as gaseous contrast was developed for direct measurement of lung respiratory motion. One healthy subject was imaged at the end-of-inhalation (EOI) and end-of-exhalation (EOE) phases in a single scan using the HP He-3 MR tagging technique, as well as a high-resolution 3D proton MR sequence (TrueFISP) which shows rich internal lung features (pulmonary vessels). 3D displacement vectors of the lungs between the EOI and EOE phases were determined via two methods: 1) by tracking the movements of the tags between the two phases in the HP He-3 MR tagging images; and 2) by registering the proton MR images at two phases using deformable image registration (DIR) via commercial software. DIR-derived 3D displacement vectors were down-sampled to approximate the resolutions of those derived from the MR tagging method. Furthermore, 3D lung ventilation maps, where ventilation is defined as (Volume_EOE-Volume_EOI)/Volume_EOI, were created based on the displacement vectors and compared between the two methods.
Results: The HP He-3 tagging MR images clearly revealed 3D lung respiratory motion between the EOI and EOE phases. Tags in the EOE phase were slightly blurred but traceable. 3D displacement vectors derived from the tagging and DIR methods showed distinct differences in motion pattern, amplitude and direction, for the entire lungs and on regional basis. Lung ventilation maps derived from two methods showed different mean values and uniformity: the mean (±SD) ventilation was -0.49 (±0.19) and 0.14 (±0.32) for the MR tagging and DIR methods, respectively.
Conclusions: Discrepancies were found in lung respiratory motion and ventilation between the HP He-3 MR tagging method and the DIR method. Thorough evaluation of DIR accuracy for the lungs is highly desired in future studies.