Validation of the Monte Carlo-based simulation tool PET-SORTEO

PET-SORTEO: a Monte Carlo based simulator to enable fast generation of realistic biological PET data

The Team

Frederic Boisson, David Zahra, Marie-Claude Gregoire and Anthonin Reilhac.
ANSTO LifeSciences

What is PET-SORTEO?

PET-SORTEO is a Monte Carlo-based based simulation tool that is able to generate realistic data in accordance with the numerical phantom description. It also takes into account the scanner geometry and physical characteristics and takes into account most of the  phenomena that alter the final image quality such as scatter and random events, and dead-time). 

One of the main strengths of the PET-SORTEO simulator lies in its ability to generate realistic data for both clinical and preclinical scanners. It has been thoroughly validated and used for the geometry of the Ecat Exact+ human scanner  ([1,3]) and for the Siemens  R4 and Focus220 preclinical scanners. A full description of the simulation model and their related parameters is given in the paper by Reilhac et al ([1]) and in Lartizien et al.

Why use Simulation?

In order to address the increasing need for simulation models of animal PET  imaging, we adapted and configured PET-SORTEO for the Inveon small-animal  PET scanner manufactured by Siemens Preclinical Solutions.

To obtain realistic data, a thorough validation of the simulator for the desired detection geometry is required. Once validated, the simulator, associated with the use of phantoms, allows statistical analysis of biological processes involved in many diseases. The simulation tool also eliminates uncertainty (bias) in tracing the regions of interest. Using segmented digital volumes provides exact knowledge of all the constituent regions of interest. Variations in the structure under study, which are almost impossible to detect in practice due to partial volume effects, are revealed.

What we are doing?

This work aimed at validating the use of the PET-SORTEO Monte Carlo simulation tool to produce accurate simulated biological studies for the Inveon small animal PET scanner. This simulation platform helped to optimize processing from acquisition, to corrections, reconstruction and statistical analysis.  In turn, this improves the detectability of biological phenomena with PET.

The outcomes

PET-SORTEO Validation:

A study presenting the  validation of SORTEO against the Inveon PET geometry has been submitted to The Journal of Nuclear Medicine. This validation was carried  out against actual measurements performed on the scanner at the Australian Nuclear  Science and Technology Organisation in Australia.

The validation included sensitivity, spatial resolution, scatter fraction, count rates, data correction and final image quality, following the NEMA NU-4 standards.

Figure 1. Experimental  and simulated results obtained using the NU-4 2008 Image Quality Phantom and the associated Recovery coefficient measurements.

Simulated biological study:

A simulated biological study was performed using a numerical mouse phantom (Digimouse). Two groups of 9 11C-raclopride scans were simulated for the mouse: a control group and  group with 10% decrease in receptor density in the striata. Raw data were reconstructed using Filtered Back Projection (FBP) with all the corrections applied. Receptor density maps were generated from each simulated experiment using the Partial Saturation Approach (PSA) (link to QuantificationPSA). The detectability of the receptor density variation between the two groups is computed as the ratio of the voxels found to be statistically different (paired t test, p < 0.05) in the striata with the total number of voxels in the same region.

The simulated study provided the detectability (80% with n=9 per group) obtained with standard processing methods in Raclopride experiments and set the framework for optimisation strategies.

Fig.2: (a) Coronal slice of a simulated Raclopride scan, (b) and (c) zooms of the brain extracted from a transverse and coronal slice, respectively. (d) Graph showing the detectability of receptor density variation in a Raclopride simulated experiment as a function of the number scans included in each group.

The Future

The validation of the PET-SORTEO Monte Carlo simulation tool for the Inveon PET scanner geometry and the associated processing line will help to improve the detectability of biological phenomena with PET.

PET-SORTEO will be used to characterize and optimize dual-mice PET acquisitions within an ANSTO LifeSciences’ strategic internal project. The objective is to ensure accurate quantification by taking into account all the variations and possible problems induced when imaging two animal simultaneously.

The simulation platform will also be used to generate an online database which will include several animal disease models. ANSTO LifeSciences aims to provide the scientific community with an online simulation platform that will enable investigations of biological phenomena. The ultimate goal is to increase the number of studies on understanding the biological processes responsible for diseases and propose effective therapies that will benefit everyone.

Publications

Validation of PET-SORTEO Monte Carlo  simulations for the geometry of the Inveon PET  preclinical scanner
Boisson, F., Wimberley, C., Zahra, D., Lehnert, W.,  Gregoire, M-C., Reilhac, A.

Project Contacts

Frederic Boisson
Anthonin Reilhac

References

1. Reilhac et al 2004
2. Reilhac et al TNS 2005
3. Reilhac et al TMI 2006