"SPECT/CT and PET/CT have had tremendous success in recent years, but CT doesn't give good soft-tissue differentiation and it adds to the ionizing radiation burden," explained Tsui, a professor of radiology at Johns Hopkins University (Baltimore, MD). "PET/MR and SPECT/MR can potentially provide additional and improved multimodality information."

Tsui noted that the first commercial simultaneous PET/MR system was unveiled by Siemens Healthcare at last year's RSNA conference. Simultaneous SPECT/MR development, however, is lagging behind. One factor slowing its commercial implementation could be the fact that SPECT systems today mostly use a scintillation detector coupled to a set of photomultiplier tubes. However, the large effect that magnetic fields have on electrons travelling through the photomultiplier tube makes this design unsuitable for use within an MRI scanner.

Instead, Tsui and colleagues – also from Gamma Medica-Ideas of Northridge, CA – are using an MR-compatible solid-state detector: cadmium zinc telluride (CZT) coupled to an ASIC. This detector works via direct conversion of incoming gamma photons into electron-hole pairs, and is minimally affected by the presence of a strong magnetic field.

System verification

Tsui and co-workers have designed and built a first-generation MR-compatible SPECT prototype comprising three rings of eight high-efficiency CZT detector modules, each module containing a 16 x 16 array of 1.6 mm CZT pixels. An outer diameter of 12 cm enables the device to fit into a small-bore small-animal MRI scanner. The 24 views provided by the detectors offer the potential for performing fast dynamic studies without the need for collimator rotation.

The researchers have also designed a cylindrical multi-pinhole collimator with 24 pinholes (to match the 24 CZT detector modules), which provides projections from a common field-of-view of around 25 mm. Due to size constraint, the system's RF coil is situated outside of the collimator.

The prototype was evaluated both in standalone mode and within a clinical 3T MRI. There was minimal interference seen on simultaneously acquired SPECT and MR images. Minimizing ferromagnetic material in the CZT modules ensured that there were no noticeable artefacts on the MR images due to the presence of the SPECT hardware.

The static magnetic field did affect the SPECT system, generating a Lorenz shift that shifted the image by approximately 1.5 mm in a 3T field. However, Tsui pointed out that this shift can easily be corrected to improve the final SPECT image quality.

The team has also developed a sparse-view image reconstruction method (based on modelling of the point response function for each pinhole), which enables creation of artefact-free SPECT images. They used the technique to reconstruct a SPECT image of a whole mouse, taken using 24 views without rotation.

Second generation

The first MR-compatible SPECT prototype demonstrated the feasibility of simultaneous SPECT/MR imaging using a multiple detector ring system. However, the system exhibited some limitations, such as relatively poor spatial resolution (3–5 mm).

To address some of these issues, Tsui and colleagues have devised a second-generation SPECT/MR prototype comprising five rings of 19 CZT detectors (each containing a 16 x 16 array of 1.6 mm pixels) and a variety of cylindrical multipinhole collimators. The new device has an outer diameter of 20 cm, enabling it to fit inside a mid-sized preclinical MRI scanner.

The larger detector rings should improve the system resolution, as well as enable the use of optimized multi-pinhole collimator designs, which in turn may improve the resolution further. The researchers examined two multi-pinhole collimator designs for mice, both offering a 30 mm field-of-view.

An 18-pinhole collimator (three rows of six pinhole inserts) offered the higher potential resolution of around 1 mm. However, this design would require rotation of the collimator during data acquisition. A 36-pinhole design, with three rows of 12 pinhole inserts, offered a slightly lower resolution of 1.5 mm, but can create artefact-free images without rotation.

Another restriction of the first-generation design is the suboptimal RF coil configuration. The larger detector ring used in the second-generation device means that there's now room for the RF coil to fit inside the collimator. Placement closer to the subject being imaged should increase the signal-to-noise ratio of the MR image.

Tsui concluded by noting that his team has just received the second-generation MR-compatible SPECT prototype from Gamma Medica-Ideas and initial testing has begun. Full evaluation of the SPECT prototype is now underway, including sample SPECT/MR imaging of a mouse kidney.