Paving the way towards clinical PET-based in-vivo treatment verification at HIT

Reporter: Gita Suneja, MD
The Abramson Cancer Center of the University of Pennsylvania
Last Modified: May 13, 2011

Presenter: Julia Bauer, PhD
Presenter’s Affiliation: Heidelburg Ion-Beam Therapy Center


  • Treatment planning generates a dose distribution that is approved by a physician and reviewed by a physicist.
  • Unfortunately, in vivo activity may not reflect this ideal dose distribution due to multiple factors.
  • Similarly, planned delivered dose and resultant in vivo activity are not identical.
  • Activity distribution is correlated to dose distribution, but not in a straightforward way. Therefore, we cannot compare activity patterns to the pre-delivery dose distribution but rather need to compare it to a model of anticipated dose delivery.
  • In vivo activity can be measured by post-irradiation PET.
  • β-emitters are formed as a by-product of irradiation in nuclear interactions between ions and irradiated tissue.
  • The subsequent pattern of beta activity can be detected by PET and can be compared with models designed by Monte Carlo simulation. This technique offers an opportunity to detect beta activity non-invasively and to immediately verify dose activity after a radiation treatment.
  • Several models for PET detection of activity exist:
    • At PSI, in-beam PET was used to measure activity during treatment and shortly thereafter. The measured PET image is compared to the calculated PET image.
    • At MGH, offline PET/CT was used to measure activity in a room separate from treatment room. The distance and time for transfer have important implications for resultant PET. Distortion must be anticipated, and washout needs to be taken into consideration.
  • In the Heidelberg Ion-Beam Therapy Center, greater than 400 patients have been treated since November 2009. A commercial full ring PET/CT scanner is installed next to the treatment rooms with an offline configuration, similar to MGH.

Materials and Methods

  • PET is able to measure the activity distribution in vivo and compare it to a model derived from the prescribed plan dose, time course of radiation, and subsequent imaging.
  • This modeling is done by Monte-Carlo simulation based on the FLUKA code. FLUKA is a general-purpose tool for calculations of particle transport and interactions with matter.
  • The models were developed by comparing radiation-induced activity in homogeneous phantoms of different composition.
  • Graphite, PE, gelatine, and PMMA were used as phantoms for model derivations. Main isotopes of interest were C-11 (t ½ 10 min), N-13 (t ½ 10 min), and O-15 (t ½ 2 min).
  • The spatial distribution is combined with a time structure (irradiation time, pauses, PET frame duration) to yield a PET model which be compared to the obtained PET image post-treatment.
  • The various phantoms were tested for carbon ions and proton beam irradiation.
  • Repeat applications of treatment at various energy levels helped to fine tune the models.
  • A graphical user interface (GUI) was also developed for integration into clinical work flow.


  • For PE, water, and PMMA phantoms, the distributions were consistent and the subsequent modeling is quite accurate. The measurements are reproducible by in repeat simulations.
  • The integration of the PET into clinical work flow is as follows: treatment planning à RTstruct à RT dose à simulation study à calculation of forward dose and beta emitter yield from PET. This model is used for future comparisons to measured PET data. PET measurements can be obtained after each fraction.
  • Challenges included appropriate delineation of the time structure to obtain activity distribution, washout modeling, registration of multiple PET-CTs, and a low signal to noise ratio.
  • This model allows for adaptation of Monte Carlo simulation to handle facility-specific, patient-specific, and plan-specific information to automatically calculate dose and activity.
  • The GUI is useful for management, visualization, data exploration, and analysis.
  • Examples of patients with GBM and H&N were provided. The difference in dose distribution calculated at time of treatment planning and in-vivo activity measured by PET is clear.

Author's Conclusions

  • This is the first PET-based in vivo treatment verification performed at HIT.
  • Extensive experimental validation of the Monte Carlo simulation for protons and carbon 12 is needed.
  • Automated Monte Carlo calculations of dose and activity are achievable. A GUI can be used to make this data accessible and available for research.

Clinical Implications

  • Dose verification is a challenge and current methods do not actually reflect in vivo activity. This work enables clinical application of PET-based post-irradiation verification of scanned ion treatments.
  • PET-based verification may provide a reliable way of estimating dose delivered to the patient and tumor, taking into account patient and tumor-specific biology and metabolism.
  • Models of dose delivery and activity may ultimately lead to changes in treatment planning. Perhaps more individualized treatment plans and dose constraints can be developed based on parameters seen at the time of PET.
  • Other areas of future improvement include improvement of washout modeling.
  • Overall, the results presented here and by other groups demonstrate the potential of PET as a very exciting modality for understanding in vivo dose delivery, the future implications of which may be quite extensive.

Frequently Asked Questions