MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Evaluating a Model-Based 4DCT Technique for Managing Respiration-Induced Motion of Abdominal Tumors

D O'Connell*, M Lauria , J Lewis , A Santhanam , A Raldow , P Lee , D Low , Department of Radiation Oncology, University of California, Los Angeles, Los Angeles, CA

Presentations

(Sunday, 7/14/2019) 3:30 PM - 4:00 PM

Room: Exhibit Hall | Forum 5

Purpose: Model-based four-dimensional computed tomography (4DCT) has been previously demonstrated to produce accurate, artifact-free breathing-gated thoracic CT images for treatment planning. In this work, we investigate the technique’s performance when applied to abdominal imaging by assessing two components: deformable image registration (DIR) and the tissue motion model.

Methods: One patient underwent imaging for a model-based 4DCT technique consisting of 25 free-breathing CT scans with simultaneous breathing surrogate monitoring. Tissue motion between scans was measured using DIR. Two DIR algorithms, ‘deeds’ and ‘pTVreg’ were used. The distance discordance metric (DDM), a previously published method to quantify spatial uncertainties in DIR, was calculated for both algorithms. Model parameters were fit for both sets of DIR results and the residuals (defined as the differences between measured and predicted tissue positions) were computed.

Results: The deeds algorithm achieved a median DDM of 1.54 ± 1.66 mm, with a 95th percentile value of 4.68 mm. The resulting model residual median was 0.84 ± 0.75 mm, with a 95th percentile of 2.51. The pTVreg achieved a median DDM of 1.40 ± 2.57 mm, with a 95th percentile value of 7.58. Resulting model residual median was 0.80 ± 1.39 mm, with a 95th percentile of 3.93. DDM correlated strongly with model residual for both algorithms (R = 0.90 and 0.81 respectively), but the DDM values for the two algorithms did not correlate well with each other (R = 0.39).

Conclusion: Two DIR algorithms were examined as part of a model-based 4DCT technique for abdominal imaging. Although the spatial distributions of uncertainty varied, both algorithms produced sufficiently consistent results to allow for motion model fitting. DDM values and model residual results suggest that both algorithms as well as a motion model initially designed for lung applications are sufficiently accurate to characterize bulk organ motion in the abdomen.

Keywords

Motion Artifacts, Respiration, CT

Taxonomy

IM- CT: 4DCT

Contact Email