Project Details
Projekt Print View

4D cone beam computed tomography as a novel tool for accurate time-resolved dose calculation in particle therapy

Subject Area Medical Physics, Biomedical Technology
Term from 2018 to 2023
Project identifier Deutsche Forschungsgemeinschaft (DFG) - Project number 399148265
 
Final Report Year 2024

Final Report Abstract

Cone beam computed tomography (CBCT) scanners are increasingly installed in state-of-theart proton therapy facilities. They employ flat panel detector technology and allow the acquisition of a three-dimensional volume in a single gantry rotation. They are currently used in imageguided photon and proton therapy to correct inter-fractional patient positioning errors. Several international research groups, including us, have developed the ability to compute proton dose distributions on intensity corrected CBCT images. This task is non-trivial due to the lower image quality of CBCT compared to diagnostic CT. CBCT based dose calculations can be used to detect errors in dose delivery caused by inter-fractional changes such as weight loss. This supports the implementation of adaptive radiation therapy (ART) where corrective measures are applied during the treatment course to restore the initial plan quality. Proton therapy is currently tackling the additional challenges raised by tumours with respiratory intra-fractional movement located in the abdominothoracic region. A four-dimensional CT scan (4DCT), which captures the patient’s average breathing cycle, is used for treatment planning. The accuracy of the resulting treatment plan for moving targets can easily be degraded by changes in the breathing pattern. The high sensitivity of protons to such changes suggests that 4D imaging at the treatment site is desirable, since it would allow ART for moving targets. With their slow rotation speed, clinical CBCT scanners intrinsically allow 4D image reconstruction by tracking the motion of the patient’s diaphragm. However, without intensity correction, the image quality of 4DCBCT is insufficient for proton dose calculation due to undersampling and scatter artefacts. In our project, we have developed three methods to correct 4DCBCT intensity and reconstruct images of high quality, despite the intrinsic issues related to sparse projections when reconstructing up to 10 breathing phases. The first method used deformable image registration (DIR) to generate a so-called 4DvCT (4DCT to 4DCBCT DIR). Subsequently, the 4DvCT was used in a projection scatter estimation pipeline which allowed projection correction followed by reconstruction, yielding a scatter corrected 4DCBCT (4DCBCTcor). Finally, to accelerate the correction procedure, we made use of artificial intelligence by training our ScatterNet model for 4DCBCTcor. The CBCT correction methods were extensively validated using a realistic porcine lung phantom allowing ground truth determination by 4DCT scanning, as well as in a retrospective cohort of lung cancer patients with 4DCT and 4DCBCT scans. Validation relied on proton therapy dose calculation accuracy and was found valuable in future lung proton therapy ART decision making. The use of artificial intelligence reduced the computational effort for 4DCBCTcor scatter correction to clinically acceptable times of few seconds.

Publications

 
 

Additional Information

Textvergrößerung und Kontrastanpassung