Filtered backprojection proton CT reconstruction along most likely paths. Issue 3 (28th February 2013)

• Record Type: Journal Article
• Authors: Rit, Simon; Dedes, George; Freud, Nicolas; Sarrut, David; Létang, Jean Michel
• Is Part Of: Medical physics. Volume 40:Issue 3(2013)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2013-02-28
• Subjects: Conformal radiation treatment -- Monte Carlo methods -- Spatial resolution -- Reconstruction -- Computed tomography; computerised tomography -- image reconstruction -- image resolution -- medical image processing -- Monte Carlo methods -- phantoms -- radiation therapy; proton computed tomography -- proton CT -- filtered backprojection -- most likely path; Computerised tomographs -- Radiation therapy -- Digital computing or data processing equipment or methods, specially adapted for specific applications -- Image data processing or generation, in general; Protons -- Spatial resolution -- Medical imaging -- Image reconstruction -- Computed tomography -- Medical image reconstruction -- Radiography -- Collisional energy loss -- Medical image spatial resolution -- Photons; Medical physics -- Periodicals; Medical physics; Geneeskunde; Natuurkunde; Toepassingen; Biophysics; Periodicals; Periodicals; Electronic journals; 610.153
• DOI: 10.1118/1.4789589
• Languages: English
• ISSNs: 0094-2405

Abstract

Abstract : Purpose: : Proton CT (pCT) has the potential to accurately measure the electron density map of tissues at low doses but the spatial resolution is prohibitive if the curved paths of protons in matter is not accounted for. The authors propose to account for an estimate of the most likely path of protons in a filtered backprojection (FBP) reconstruction algorithm. Methods: : The energy loss of protons is first binned in several proton radiographs at different distances to the proton source to exploit the depth‐dependency of the estimate of the most likely path. This process is named the distance‐driven binning. A voxel‐specific backprojection is then used to select the adequate radiograph in the distance‐driven binning in order to propagate in the pCT image the best achievable spatial resolution in proton radiographs. The improvement in spatial resolution is demonstrated using Monte Carlo simulations of resolution phantoms. Results: : The spatial resolution in the distance‐driven binning depended on the distance of the objects from the source and was optimal in the binned radiograph corresponding to that distance. The spatial resolution in the reconstructed pCT images decreased with the depth in the scanned object but it was always better than previous FBP algorithms assuming straight line paths. In a water cylinder with 20 cm diameter, the observed range of spatial resolutions was 0.7 − 1.6 mm compared to 1.0 − 2.4 mm at best with a straight line path assumption. The improvement was strongly enhanced in shorter 200° scans. Conclusions: : Improved spatial resolution was obtained in pCT images with filtered backprojection reconstruction using most likely path estimates of protons. The improvement in spatial resolution combined with the practicality of FBP algorithms compared to iterative reconstruction algorithms makes this new algorithm a candidate of choice for clinical pCT.