Engineering innovation leads to first clinical PET/MR scanner

October 10, 2005

Working with colleagues at the former CTI PET Systems in Knoxville, Siemens Medical Solutions engineers in Erlangen, Germany, are assembling a prototype PET/MR scanner designed to overcome technical barriers that have thus far kept a hybrid of these two modalities from clinical use.

Working with colleagues at the former CTI PET Systems in Knoxville, Siemens Medical Solutions engineers in Erlangen, Germany, are assembling a prototype PET/MR scanner designed to overcome technical barriers that have thus far kept a hybrid of these two modalities from clinical use.

The device will be equipped with a tubular PET insert, featuring avalanche photo diode (APD) detectors in lieu of photomultiplier tubes, inserted in the bore of a commercially available 3T MR scanner. PET and MR images will be acquired simultaneously, delivering a combination of anatomic localization and physiologic measurements similar to that of fusion imaging with PET/CT, according to Matthias J. Schmand, Ph.D., director of detector R&D for Siemens

Siemens engineers discussed their R&D efforts at the Society of Molecular Imaging meeting in September. A plenary session at the conference in Cologne, Germany, also covered the 10-year effort by industry and academic researchers to design a workable PET/MR scanner for clinical and research applications.

Siemens' prototype and the first clinical scanner based on that platform will be devoted to neurologic applications, according to lead project engineer Ralf Ladebeck. The limited field-of-view possible with the current PET/MRI insert makes the device most effective in such applications. The prototype will be completed in 2006.

Many of the same goals that led to PET/CT's development have spurred the quest for a workable PET/MR system, according to Bernd Pichler, Ph.D., a medical physicist at the University of Tuebingen in Germany. Both hybrid approaches fuse physiologic data acquired with FDG-PET and anatomic information captured with either CT or MRI to localize metabolic activity.

But because MRI does not involve ionizing radiation, PET/MR may be preferable to PET/CT for pediatric applications and patients requiring repeat imaging. The high-contrast resolution of MRI could make its pairing with PET the first choice of brain studies, stem cell trafficking, and other studies which requires clinician confidence in the modality's characterization of soft tissue, Pichler said.

Radiologists would also like to explore the diagnostic potential of combined PET and functional MR studies. The ability to simultaneously acquire FDG-PET with MR spectroscopy, blood oxygenation, or diffusion tensor imaging could improve the characterization of multiple sclerosis, Parkinson's disease, Alzheimer's disease, and other brain disorders.

To realize these goals, engineers had to decide which of several possible scanner configurations is best suited for hybrid PET/MR imaging. They considered using a sequential arrangement comparable to PET/CT, with a PET scanner set behind an MR scanner on a single patient bed. This approach is easy to execute, but simultaneous data acquisition cannot be performed, Ladebeck said.

A second option involved developing a PET insert for an existing MR scanner. Simultaneous imaging is possible by inserting the detector against the inner rim of the MR patient bore, but this presented numerous technical problems, Ladebeck said. Foremost among these challenges was how to design a detector system without using ferrous material that would interfere with the homogeneity of the MR scanner's magnetic field or produce eddy currents or vibrations from its mechanical components.

"We don't want to make a poor MR to get a high-end PET scanner or the other way around," he said. "We would like to go for the best of both components."

Miniaturization was also essential to avoid sacrificing too much of the bore diameter to the additional components.

Simon Cherry, Ph.D., now a professor of biomedical engineering at the University of California, Davis, laid the groundwork for this design, according to Pichler. Mouse PET/MR scanner development work at Cherry's laboratory at UCD demonstrated in 2004 that in vivo PET imaging using APD detectors is feasible. Pichler and colleagues at the University of Munich and University of Tuebingen also played seminal roles.

The Siemens prototype will use a Magnetom Trio 3T system and head coil as the platform for the combined PET/MR scanner.

The PET insert pairs APD detectors in a block configuration with lutetium oxyorthosilicate crystal scintillators to maximize their energy resolution within the confined space, Schmand said. LSO features a better light yield than bismuth germanate or gadolinium silicate detectors, producing a cost-effective block detector design for positioning and timing, Schmand said. Crystals are arranged in an eight-by-eight crystal configuration coupled to a two-by-two APD array creating a 64:4 ratio of LSO crystals to APD sensor channels. Four-detector blocks are assembled in 32 x 32 x 20-mm³ detector modules. Proprietary application-specific integrated circuits supply preamplification.

Overall, the transaxial depth of the detector modules is 5 cm, reducing the width of the MR bore by 10 cm. With such a flat-detector profile, Siemens officials envision the possibility of a fully integrated 60-cm open-bore whole-body oncology PET/MR using the firm's 70-cm-bore Magnetom Espree. The prototype PET/MR insert will have an opening of about 35 cm.

Phantom testing of an experimental configuration of the scanner produced a 3.2-cm diameter field-of-view. Structures as small as 2 mm could be resolved with the PET detector, and objects as small as 1 mm could be seen with a host MR scanner. The signal-to-noise ratio of the MRI equipped with the insert was about 15% lower than with the component removed. The SNR loss for the PET component was 20%, Schmand said.

Siemens engineers are aiming at achieving 2.5-mm isotropic resolution with the prototype system, Schmand said.

"The goal is to develop a system so we have no compromises," Schmand said.