(Sunday, 7/14/2019) 3:30 PM - 4:00 PM
Room: Exhibit Hall | Forum 5
Purpose: 4DCT is the standard to assess thoracic and abdominal tumor motion. Using 4DCT to delineate the internal target volume (ITV) assumes a regular breathing that is not true for many patients. We developed a modeling technique to derive dynamic CT image series from 4DCT for patients with irregular breathing and compared ITVs from 4DCT and dynamic CT.
Methods: Dynamic CT consists of a time series of 3DCT created by deforming a 3D reference CT using time-dependent deformation maps. In this study, we first selected the most stable phase in 4DCT (smallest variation in displacement) as the reference CT. We used deformable image registration and principal component analysis to calculate two major modes of deformations. Then 4DCT was broken down to a series of 2D axial images at different acquisition times. The deformation maps at these acquisition times linearly combined the major modes of deformation with coefficients determined by optimally correlating the dynamic CT with the 2D axial images. ITV was derived by deforming the gross tumor volume (GTV) drawn on the reference phase using the time-variant deformation maps. For a lung and a liver patient with irregular breathing motion during 4DCT acquisition, we reconstructed the dynamic CTs and the ITV volumes. The ITVs from the dynamic CTs were compared with the ITVs manually drawn from the 4DCT.
Results: For the lung patient, ITV from the dynamic CT was 0.8 cm larger in size and 4.3 cc (34%) greater in volume when compared with ITV from the 4DCT. For the liver patient, ITV from the dynamic CT was 0.9 cm larger in size and 64 cc (17%) greater in volume.
Conclusion: Motion modeling can be used to derive dynamic CT image series and generate ITV volumes for patients with irregular breathing. 4DCT underestimates ITV for patients with irregular breathing.
Keywords
Motion Artifacts, Modeling, CT
Taxonomy
Not Applicable / None Entered.
Contact Email
