Schematic of an idealized single-proton-tracking CT scanner. Protons with known incident energy are individually recorded in four planes of position-sensitive detectors (e.g., silicon strip detectors), forming the scanner reference system . The detectors provide positions as well as azimuth and declination angles of the protons in front and behind the object. For a complete scan, the object is traversed by broad proton beams from many different projection angles . The resulting cone-beam data set allows reconstruction of the relative electron density distribution in the object reference system . The exit energy of each proton is recorded with an energy detector (e.g., a crystal calorimeter) in coincidence with its position and angle information.
Schematic cross section of the phantom used to study the performance of the idealized proton CT scanner shown in Fig. 1. The cylindrical water phantom contains three types of tissue inserts [compact bone (ICRU), adipose tissue (ICRP), and skeletal muscle (ICRU)] grouped into three cylinders of 3, 10, and diameter, respectively. The electron densities relative to water are listed for each tissue type and were derived from the elemental composition of each tissue. (Ref. 12).
Standard deviation of the energy loss distribution as a function of water thickness traversed by protons. The solid lines correspond to analytical calculations using Bohr’s and Tschalar’s theories of energy straggling, respectively, while the points correspond to the standard deviations of the GEANT4-simulated energy loss distributions. The error bars correspond to the uncertainty of the standard deviations of the simulated distributions.
Relative electron density resolution in the center of a cylindrical phantom as a function of phantom diameter in centimeters and proton energy. The sampling interval is . The initial proton fluence was chosen to deliver a dose of at the center of the phantom.
Relative electron density resolution as a function of dose at the center of a -diam cylindrical phantom for protons of 175, 200, and predicted by Eq. (13). For comparison, the dose-resolution relationship for an ideal x-ray CT scanner with photons is also shown. The sampling interval is for both cases. The discrete points represent the noise levels of the reconstructed proton CT images shown in Fig. 6 at three different dose levels. Circular symbols correspond to images reconstructed excluding secondary protons, and triangular symbols to reconstructed images including secondary protons.
Reconstructed proton CT images based on GEANT4-simulated proton CT data of the cylindrical water phantom shown in Fig. 2. Reconstructions were completed for three different proton doses at the center of the phantom (3.1, 1.6, and ). For each of 180 projections taken at 2° intervals, proton data were binned into 240 intervals over a distance of , corresponding to a sampling interval of . Nearest-neighbor mapping was used to project average bin values onto a regular map in sinogram space. For reconstruction, the filtered backprojection algorithm was applied using a Ram-Lak filter with a cutoff frequency of . The images on the left were reconstructed including secondary protons, and the images on the right were reconstructed excluding secondary protons.
Article metrics loading...
Full text loading...