[Paper Review] Analysis of time-profiles with in-beam PET monitoring in charged particle therapy

A. Kraan, S. Muraro+15 more|arXiv (Cornell University)|Sep 28, 2018

Radiation Therapy and Dosimetry19 references3 citations

TL;DR

This study evaluates in-beam PET monitoring in charged particle therapy by analyzing time-profiles of positron-emitting isotopes (11C, 15O, 10C) shortly after irradiation. Using FLUKA Monte Carlo simulations and experimental data from phantoms irradiated with 130 MeV protons, it demonstrates good agreement between measured and simulated spatial and decay rate profiles, with a notable excess of 10C in data compared to MC predictions—particularly in high-density polyethylene phantoms—highlighting potential limitations in nuclear interaction models for short-lived isotopes.

ABSTRACT

Background: Treatment verification with PET imaging in charged particle therapy is conventionally done by comparing measurements of spatial distributions with Monte Carlo (MC) predictions. However, decay curves can provide additional independent information about the treatment and the irradiated tissue. Most studies performed so far focus on long time intervals. Here we investigate the reliability of MC predictions of space and time (decay rate) profiles shortly after irradiation, and we show how the decay rates can give an indication about the elements of which the phantom is made up. Methods and Materials: Various phantoms were irradiated in clinical and near-clinical conditions at the Cyclotron Centre of the Bronowice proton therapy centre. PET data were acquired with a planar 16x16 cm$^2$ PET system. MC simulations of particle interactions and photon propagation in the phantoms were performed using the FLUKA code. The analysis included a comparison between experimental data and MC simulations of space and time profiles, as well as a fitting procedure to obtain the various isotope contributions in the phantoms. Results and conclusions: There was a good agreement between data and MC predictions in 1-dimensional space and decay rate distributions. The fractions of $^{11}$C, $^{15}$O and $^{10}$C that were obtained by fitting the decay rates with multiple simple exponentials generally agreed well with the MC expectations. We found a small excess of $^{10}$C in data compared to what was predicted in MC, which was clear especially in the PE phantom.

Motivation & Objective

• To assess the reliability of FLUKA Monte Carlo simulations in predicting spatial and temporal decay profiles of positron-emitting isotopes (11C, 15O, 10C) in phantoms shortly after proton irradiation.
• To extract relative isotope fractions from experimental decay curves using multi-exponential fitting, enabling tissue composition inference.
• To investigate discrepancies between measured and simulated 10C contributions, particularly in carbon-rich materials like high-density polyethylene.
• To demonstrate the feasibility of mapping isotope distributions (15O, 11C, 10C) in space and time using short-time interval PET data for future biological and dosimetric applications.

Proposed method

• Irradiated three homogeneous phantoms (PMMA, high-density PE, water) and one inhomogeneous Zebra phantom (PMMA/PE layers) with 130 MeV proton pencil beams (10^10 protons, 5 s exposure).
• Acquired time-resolved PET data using a planar 16×16 cm² LYSO-based system (DoPET) with a non-conventional Straightforward Reconstruction Approach (SRA) for simultaneous spatial and temporal event reconstruction.
• Performed FLUKA simulations of particle transport and photon propagation to predict activity distributions in space and time, including isotope-specific decay contributions.
• Fitted experimental decay curves (8–300 s post-irradiation) with a sum of three exponential functions to extract relative fractions of 15O (t₁/₂=2 min), 11C (t₁/₂=20 min), and 10C (t₁/₂=19 s).
• Compared measured and simulated 1D z-profiles and decay rates across phantoms, and generated spatial maps of isotope contributions by slicing the phantom in 2 mm z-steps.
• Validated that neglected isotopes (e.g., 5B, 14O, 13N) contributed less than 2% to the total activity in the selected time window.

Experimental results

Research questions

• RQ1 How accurately do FLUKA simulations predict the spatial and temporal decay profiles of 11C, 15O, and 10C in phantoms irradiated with 130 MeV protons?
• RQ2 To what extent do measured decay rates in the first 5 minutes post-irradiation reflect the expected isotope fractions from Monte Carlo simulations?
• RQ3 Why is there a persistent excess of 10C in experimental data compared to FLUKA predictions, especially in high-density polyethylene?
• RQ4 Can time-resolved PET data be used to map the spatial distribution of different positron-emitting isotopes (15O, 11C, 10C) in heterogeneous phantoms?
• RQ5 What implications does the observed 10C discrepancy have for tissue composition analysis and biological modeling in in-beam PET?

Key findings

• Good agreement was observed between measured and FLUKA-predicted 1D z-profiles of activity, with the Zebra phantom’s layered structure clearly resolved due to differential 15O production in PMMA vs. PE.
• The relative fraction of 10C was consistently higher in experimental data than in FLUKA simulations, with a 20.7% data fraction vs. 14.3% MC prediction in high-density PE phantoms.
• In water and PMMA phantoms, the 10C fraction was also elevated in data (1.6% vs. 2.4% MC and 8.8% vs. 6.6% MC, respectively), though less pronounced than in PE.
• The 15O fraction was well-predicted by FLUKA (89.4% data vs. 91.5% MC in water), confirming accurate modeling of oxygen-rich tissue interactions.
• Spatial mapping of isotopes revealed distinct patterns: 15O was abundant in PMMA regions but absent in PE, while 10C was uniformly distributed but decayed rapidly, consistent with its short half-life.
• The 10C map in data showed a discrepancy with MC simulations, especially deeper in the phantom, suggesting potential limitations in modeling short-lived isotope production or transport in dense materials.