On proton CT reconstruction using MVCT-converted virtual proton projections

Dongxu Wang, T. Rockwell MacKie, Wolfgang A. Tome

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

Purpose: To describe a novel methodology of converting megavoltage x-ray projections into virtual proton projections that are otherwise missing due to the proton range limit. These converted virtual proton projections can be used in the reconstruction of proton computed tomography (pCT). Methods: Relations exist between proton projections and multispectral megavoltage x-ray projections for human tissue. Based on these relations, these tissues can be categorized into: (a) adipose tissue; (b) nonadipose soft tissues; and (c) bone. These three tissue categories can be visibly identified on a regular megavoltage x-ray computed tomography (MVCT) image. With an MVCT image and its projection data available, the x-ray projections through heterogeneous anatomy can be converted to the corresponding proton projections using predetermined calibration curves for individual materials, aided by a coarse segmentation on the x-ray CT image. To show the feasibility of this approach, mathematical simulations were carried out. The converted proton projections, plotted on a proton sinogram, were compared to the simulated ground truth. Proton stopping power images were reconstructed using either the virtual proton projections only or a blend of physically available proton projections and virtual proton projections that make up for those missing due to the range limit. These images were compared to a reference image reconstructed from theoretically calculated proton projections. Results: The converted virtual projections had an uncertainty of ±0.8 compared to the calculated ground truth. Proton stopping power images reconstructed using a blend of converted virtual projections (48) and physically available projections (52) had an uncertainty of ±0.86 compared with that reconstructed from theoretically calculated projections. Reconstruction solely from converted virtual proton projections had an uncertainty of ±1.1 compared with that reconstructed from theoretical projections. If these images are used for treatment planning, the average proton range uncertainty is estimated to be less than 1.5 for an imaging dose in the milligray range. Conclusions: The proposed method can be used to convert x-ray projections into virtual proton projections. The converted proton projections can be blended with existing proton projections or can be used solely for pCT reconstruction, addressing the range limit problem of pCT using current therapeutic proton machines.

Original languageEnglish (US)
Pages (from-to)2997-3008
Number of pages12
JournalMedical Physics
Volume39
Issue number6
DOIs
StatePublished - Jun 2012

Fingerprint

Protons
Tomography
X-Rays
Uncertainty

Keywords

  • pCT reconstruction
  • proton projection
  • range limit
  • sinogram conversion
  • stopping power

ASJC Scopus subject areas

  • Biophysics
  • Radiology Nuclear Medicine and imaging

Cite this

On proton CT reconstruction using MVCT-converted virtual proton projections. / Wang, Dongxu; MacKie, T. Rockwell; Tome, Wolfgang A.

In: Medical Physics, Vol. 39, No. 6, 06.2012, p. 2997-3008.

Research output: Contribution to journalArticle

Wang, Dongxu ; MacKie, T. Rockwell ; Tome, Wolfgang A. / On proton CT reconstruction using MVCT-converted virtual proton projections. In: Medical Physics. 2012 ; Vol. 39, No. 6. pp. 2997-3008.
@article{c435e4b0593f4554b287de345d30fe6e,
title = "On proton CT reconstruction using MVCT-converted virtual proton projections",
abstract = "Purpose: To describe a novel methodology of converting megavoltage x-ray projections into virtual proton projections that are otherwise missing due to the proton range limit. These converted virtual proton projections can be used in the reconstruction of proton computed tomography (pCT). Methods: Relations exist between proton projections and multispectral megavoltage x-ray projections for human tissue. Based on these relations, these tissues can be categorized into: (a) adipose tissue; (b) nonadipose soft tissues; and (c) bone. These three tissue categories can be visibly identified on a regular megavoltage x-ray computed tomography (MVCT) image. With an MVCT image and its projection data available, the x-ray projections through heterogeneous anatomy can be converted to the corresponding proton projections using predetermined calibration curves for individual materials, aided by a coarse segmentation on the x-ray CT image. To show the feasibility of this approach, mathematical simulations were carried out. The converted proton projections, plotted on a proton sinogram, were compared to the simulated ground truth. Proton stopping power images were reconstructed using either the virtual proton projections only or a blend of physically available proton projections and virtual proton projections that make up for those missing due to the range limit. These images were compared to a reference image reconstructed from theoretically calculated proton projections. Results: The converted virtual projections had an uncertainty of ±0.8 compared to the calculated ground truth. Proton stopping power images reconstructed using a blend of converted virtual projections (48) and physically available projections (52) had an uncertainty of ±0.86 compared with that reconstructed from theoretically calculated projections. Reconstruction solely from converted virtual proton projections had an uncertainty of ±1.1 compared with that reconstructed from theoretical projections. If these images are used for treatment planning, the average proton range uncertainty is estimated to be less than 1.5 for an imaging dose in the milligray range. Conclusions: The proposed method can be used to convert x-ray projections into virtual proton projections. The converted proton projections can be blended with existing proton projections or can be used solely for pCT reconstruction, addressing the range limit problem of pCT using current therapeutic proton machines.",
keywords = "pCT reconstruction, proton projection, range limit, sinogram conversion, stopping power",
author = "Dongxu Wang and MacKie, {T. Rockwell} and Tome, {Wolfgang A.}",
year = "2012",
month = "6",
doi = "10.1118/1.4711752",
language = "English (US)",
volume = "39",
pages = "2997--3008",
journal = "Medical Physics",
issn = "0094-2405",
publisher = "AAPM - American Association of Physicists in Medicine",
number = "6",

}

TY - JOUR

T1 - On proton CT reconstruction using MVCT-converted virtual proton projections

AU - Wang, Dongxu

AU - MacKie, T. Rockwell

AU - Tome, Wolfgang A.

PY - 2012/6

Y1 - 2012/6

N2 - Purpose: To describe a novel methodology of converting megavoltage x-ray projections into virtual proton projections that are otherwise missing due to the proton range limit. These converted virtual proton projections can be used in the reconstruction of proton computed tomography (pCT). Methods: Relations exist between proton projections and multispectral megavoltage x-ray projections for human tissue. Based on these relations, these tissues can be categorized into: (a) adipose tissue; (b) nonadipose soft tissues; and (c) bone. These three tissue categories can be visibly identified on a regular megavoltage x-ray computed tomography (MVCT) image. With an MVCT image and its projection data available, the x-ray projections through heterogeneous anatomy can be converted to the corresponding proton projections using predetermined calibration curves for individual materials, aided by a coarse segmentation on the x-ray CT image. To show the feasibility of this approach, mathematical simulations were carried out. The converted proton projections, plotted on a proton sinogram, were compared to the simulated ground truth. Proton stopping power images were reconstructed using either the virtual proton projections only or a blend of physically available proton projections and virtual proton projections that make up for those missing due to the range limit. These images were compared to a reference image reconstructed from theoretically calculated proton projections. Results: The converted virtual projections had an uncertainty of ±0.8 compared to the calculated ground truth. Proton stopping power images reconstructed using a blend of converted virtual projections (48) and physically available projections (52) had an uncertainty of ±0.86 compared with that reconstructed from theoretically calculated projections. Reconstruction solely from converted virtual proton projections had an uncertainty of ±1.1 compared with that reconstructed from theoretical projections. If these images are used for treatment planning, the average proton range uncertainty is estimated to be less than 1.5 for an imaging dose in the milligray range. Conclusions: The proposed method can be used to convert x-ray projections into virtual proton projections. The converted proton projections can be blended with existing proton projections or can be used solely for pCT reconstruction, addressing the range limit problem of pCT using current therapeutic proton machines.

AB - Purpose: To describe a novel methodology of converting megavoltage x-ray projections into virtual proton projections that are otherwise missing due to the proton range limit. These converted virtual proton projections can be used in the reconstruction of proton computed tomography (pCT). Methods: Relations exist between proton projections and multispectral megavoltage x-ray projections for human tissue. Based on these relations, these tissues can be categorized into: (a) adipose tissue; (b) nonadipose soft tissues; and (c) bone. These three tissue categories can be visibly identified on a regular megavoltage x-ray computed tomography (MVCT) image. With an MVCT image and its projection data available, the x-ray projections through heterogeneous anatomy can be converted to the corresponding proton projections using predetermined calibration curves for individual materials, aided by a coarse segmentation on the x-ray CT image. To show the feasibility of this approach, mathematical simulations were carried out. The converted proton projections, plotted on a proton sinogram, were compared to the simulated ground truth. Proton stopping power images were reconstructed using either the virtual proton projections only or a blend of physically available proton projections and virtual proton projections that make up for those missing due to the range limit. These images were compared to a reference image reconstructed from theoretically calculated proton projections. Results: The converted virtual projections had an uncertainty of ±0.8 compared to the calculated ground truth. Proton stopping power images reconstructed using a blend of converted virtual projections (48) and physically available projections (52) had an uncertainty of ±0.86 compared with that reconstructed from theoretically calculated projections. Reconstruction solely from converted virtual proton projections had an uncertainty of ±1.1 compared with that reconstructed from theoretical projections. If these images are used for treatment planning, the average proton range uncertainty is estimated to be less than 1.5 for an imaging dose in the milligray range. Conclusions: The proposed method can be used to convert x-ray projections into virtual proton projections. The converted proton projections can be blended with existing proton projections or can be used solely for pCT reconstruction, addressing the range limit problem of pCT using current therapeutic proton machines.

KW - pCT reconstruction

KW - proton projection

KW - range limit

KW - sinogram conversion

KW - stopping power

UR - http://www.scopus.com/inward/record.url?scp=84863506804&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84863506804&partnerID=8YFLogxK

U2 - 10.1118/1.4711752

DO - 10.1118/1.4711752

M3 - Article

C2 - 22755684

AN - SCOPUS:84863506804

VL - 39

SP - 2997

EP - 3008

JO - Medical Physics

JF - Medical Physics

SN - 0094-2405

IS - 6

ER -