Image Reconstruction with a Fast, Monolithic Proton Radiography System

Abstract

Purpose/Objective(s): Margin reduction and hypo-fractionated treatment of pancreatic cancer is increasing in interest and practice in radiation oncology, but pancreatic motion has the potential to limit clinical efficacy. This study details the magnitude of observed positional variations of pancreatic tumors treated with radiation therapy utilizing the results of daily image guided radiation therapy (IGRT). Materials/Methods: Daily IGRT shifts based on cone-beam CT (CBCT) and mega-voltage CT (MVCT) were recorded and evaluated in order to estimate positioning variations of pancreatic target volumes. A total of 72 patients and 1159 3D-images captured immediately prior to daily conventionally fractionated RT were considered. Overall positioning variations, inter-patient variations, and correlations between position variations and patient-specific factors including PTV volume, tumor location (head/body/tail of pancreas), and central or lateral positioning of the PTV were estimated. Statistical analysis of the positioning variations included Kolmogorov-Smirnov (KS) testing for normality of distributions. Estimation of appropriate margins was carried out based on MZ2.5S+0.7s assuming the measured inter-fraction shifts are representative of intrafraction motion, and present a conservative estimate of spatial expansions to ensure coverage of the CTV. Results: Mean and standard deviation of IGRT shifts for all patients was 0.2AE5.2 mm in lateral directions,-0.3AE4.3 mm in vertical directions, and-0.7AE5.5 mm in longitudinal directions. The systematic variation (the standard deviation of inter-patient average shifts) was 2.6 mm, 2.6 mm, and 3.0 mm in lateral, vertical, and longitudinal directions. The random variation (the average of inter-patient standard deviations) was 3.9 mm, 3.0 mm, and 4.1 mm lateral, vertical, and longitudinal directions. Combined, these results suggest uniform margins about the CTV up to 10.4 mm are sufficient to account for observed positioning variations during IGRT. For each patient, KS testing implies the patient-specific shifts were representative of Gaussian distributions for 70/72 patients and 2/216 patient-specific directional distributions. The patient-specific distributions were not correlated with PTV volume, central or lateral location of the PTV, or whether the tumor was in the pancreatic head, tail, or body. Conclusion: Pancreas is a mobile organ with potential for large motion; our results suggest this motion is normally distributed and well-represented by Gaussian distributions. If intra-and inter-fraction motion are similar, which is plausible if pancreas motion is due to breathing and gastrointestinal motion, our data suggest margins of 10.4 mm will account for these spatial variations. These margins will surely overlap with surrounding bowel, duodenum, and/or stomach and imply dose escalation of margin-based PTVs may not be feasible.

The accuracy of helium ion CT based particle therapy range prediction: an experimental study comparing different particle and x-ray CT modalities

“…For pCT at a cyclotron facility with uninterrupted data acquisition, a planning CT scan takes approximately six minutes with the US pCT prototype (Johnson et al 2016). Ongoing efforts on detector developments (Welsh et al 2017, Esposito et al 2018, Pettersen et al 2019 aim at an acquisition time of just 1 or 2 min in the near future. In order to enable continuous data acquisition for patient scans, either a rotating chair or a gantry capable of continuous irradiation during rotation is needed.…”

An Edge-Directed Diffusion Equation-Based Image Restoration Approach for Font Generation

“…DeJongh et al used Markov network to establish the relationship model between the local image area and the basic scene, and used confidence propagation algorithm to train Markov network for image HR reconstruction. The image reconstruction effect is good, but the algorithm relies on a large number of training data sets, and its generality is poor [21], [22]. Abubakar et al proposed a sparse representation HR algorithm based on overcomplete dictionary pairs [23].…”