Click here to


Are you sure ?

Yes, do it No, cancel

Title: Interfraction Motion Modeling (IMM) Using Principal Component Analysis (PCA) for CBCT Image Estimation with Limited Projections

Z Jiang*, Y Zhang , F Yin , L Ren , Duke University Medical Center, Durham, NC


(Sunday, 7/29/2018) 5:05 PM - 6:00 PM

Room: Room 207

Purpose: To investigate the feasibility of using interfraction motion modeling (IMM) and weighted free-form deformation (WFD) to reconstruct onboard CBCT based on prior knowledge and limited projections for target localization in lung radiotherapy.

Methods: Limited-angle intrafraction verification (LIVE) system has been previously proposed to reconstruct CBCT from limited projections using prior information. CBCT is considered as a deformation of planning CT, and deformation field map (DFM) is solved by motion modeling (MM) and WFD optimization. However, the LIVE MM models respiratory motion which is not applicable for breath-hold treatments. In addition, its modeling is based on planning 4D-CT, which doesn’t reflect the day-to-day interfraction motion or deformation. In this study, we first extract interfraction motion models by applying PCA analysis on the daily DFMs registered between CT and previous days’ CBCT. CBCT on the current day is then estimated based on IMM followed by WFD using the data fidelity constraint. The method was evaluated using one breath-hold lung patient data. IMM was built based on the DFMs from the first four days, and used to estimate CBCT on the fifth day using 36 projections over 360 scan angle. The estimated CBCT was evaluated against full sampling CBCT qualitatively and quantitatively using normalized cross-correlation (NCC) of tumor region and the registered shifts between them.

Results: NCC of tumor region were 0.947 and 0.933 for IMM-WFD and WFD-only method, respectively. Tumor shifts from reconstructed CBCT to full sampling CBCT were (-0.07, -1.01, 1.10) mm and (0.17, -0.97, 1.40) mm for IMM-WFD and WFD-only method, respectively. Qualitatively, IMM-WFD estimated CBCT demonstrated a smoother tumor edge and more accurate tumor volume compared with WFD-only estimated CBCT.

Conclusion: Preliminary results demonstrated the feasibility of IMM for CBCT estimation from limited projections, which can be used for both interfraction and intrafration verification in lung radiotherapy.

Funding Support, Disclosures, and Conflict of Interest: This work is supported by NIH (Grant No.: R01 CA-184173).


Cone-beam CT, Reconstruction, Modeling


IM- Cone Beam CT: Image Reconstruction

Contact Email