Double the image, double the data
PET and MRI are combined in one instrument.
A new biomedical instrument allows a successful, simultaneous combination of MRI and PET, achieving about the same resolution and sensitivity as the individual components. It therefore opens up research and clinical possibilities.
“Combined PET/computed tomography was a major breakthrough, and I believe PET/MRI has the potential of being even more successful,” said Ciprian Catana, a member of the team that developed the device.
Now an instructor at Harvard Medical School in Boston, Catana was doing graduate work at the University of California, Davis, at the time the instrument was built there under the direction of Simon R. Cherry. Other members were from California Institute of Technology in Pasadena and from the University of Tübingen in Germany.
For biomedical imaging, PET offers significant advantages but also some substantial drawbacks. The technique permits noninvasive imaging of picomolar concentrations of radiolabeled biomolecules, but it lacks spatial resolution, and it can be difficult to tie the image to specific anatomy. On the other hand, MRI offers high spatial resolution and contrast but its molar sensitivity is orders of magnitude less than PET’s.
Researchers designed a PET insert to be placed inside an MRI scanner for simultaneous MR/PET imaging (A). A close-up view (B) of the electronics and an image (C) of the insert in place inside a preclinical MRI system also are shown. LSO = lutetium oxyorthosilicate; DAQ = data acquisition; PCB = printed circuit board; PSAPD = position-sensitive avalanche photodiode; RF = radio frequency. Images reprinted with permission from PNAS.
Because the capabilities of PET and MRI complement each other, it makes sense to combine them. Ideally, such dual scans would be done at the same time to avoid problems correlating the sets of data. Unfortunately, this is not as simple as placing one instrument inside the other and turning both on.
MRI uses strong and varying magnetic fields that can wreak havoc with typical PET detectors, which are based on photomultiplier tubes and their associated electronics. Conversely, the materials in PET electronics can disturb the carefully constructed magnetic field used in MRI.
Previous attempts to merge the two techniques have relied on long optical fibers to move the PET detectors out of the magnetic field. The resulting devices worked but were cumbersome and did not perform as well as the individual instruments did.
Cherry’s group took a different approach, using avalanche photodiodes in its system instead of photomulti-plier tubes. “The avalanche photodiode technology only relatively recently entered the PET field, and it needed some time to mature,” Catana said.
Avalanche photodiodes are insensitive to magnetic fields, allowing them to be placed in the MRI scanner. However, they have a substantially worse signal-to-noise ratio than photomultiplier tubes, particularly for low-level signals. The team worked around this problem through such design modifications as cooling the avalanche photodiodes to –10 °C, which increased sensitivity and minimized temperature drift.
The investigators constructed an array of 16 modules, with short optical fiber runs connecting lutetium oxyorthosilicate detectors to position-sensitive avalanche photodiodes from Radiation Monitoring Devices Inc. of Watertown, Mass. They mounted the photodiodes and electronics on a circuit board populated with nonmagnetic components, shielding as needed with copper laminate. They designed the entire assembly to fit inside a typical preclinical MRI scanner and to be easily inserted and removed as needed. The device’s position is adjustable so that the fields of view of the two instruments are coincident.
These images of a mouse tumor were acquired with a specially designed PET insert operating inside an MRI scanner. In A, the PET image is in the upper left, while the MR image is in the upper right. The image below shows the combined data sets. A region of hyperintensity on MRI close to the center of the tumor shows low radiotracer uptake (arrow), likely indicative of necrosis. In B, fused PET and MR images are shown from multiple transaxial sections spanning from the top of the head to the bladder. The same false-color scales were used in A and B.
They decided not to put the avalanche photodiodes in the active MRI volume, fearing increased interference between the two systems. A downside to this choice was the possibility of scintillation light loss along the optical fiber, but they minimized this by using as short a run as was feasible.
After constructing the PET insert, the researchers tested its performance alone, and they tested how a Bruker Biospin small-animal magnetic resonance system performed with the PET insert inside of it.
In the first sequence of tests, they performed phantom MRI by collecting signal-to-noise and uniformity data on a solution with known magnetic resonance characteristics and spatial distribution. As part of the same sequence, they performed similar tests for the PET scanner, examining the detector’s energy and spatial resolution. They also measured sensitivity using the magnetic resonance sequences typically employed in small-animal studies.
They found that the magnetic fields had some slight effects on the PET measurements. For example, detector maps demonstrated a small clockwise or counterclockwise rotation, depending on the detector orientation to the magnetic field. They also found a PET count rate decrease of 10 percent for one particular sequence and much less for others. There was no mispositioning of the detected events, so the spatial resolution was not affected.
Catana noted that these findings suggested that the MR pulse sequences corrupt the PET data at certain points in the sequence cycle, but only for a small fraction of the total acquisition time.
The PET insert had virtually no effect on the MRI system, which Catana said was not surprising given that the electronics were outside of the MRI viewing region.
As a final demonstration, the investigators performed in vivo mouse imaging. In one experiment, they demonstrated increased tumor uptake of a glucose analog radiotracer in an animal with a tumor roughly 9 mm in diameter. In another, they used a sodium fluoride radiotracer, moving the animal repeatedly to get a whole body scan. The dual PET and MR images showed excellent registration along the whole body axis.
Although clearly demonstrating the possibilities of the system for both imaging modalities, the researchers noted that the insert does not offer the same performance as a stand-alone PET system. For that, the insert will need a longer axial field of view and more efficient detectors. Such improvements are planned. The work was published in the March 11 issue of PNAS.
It is possible that a commercial version will be developed. In addition, clinical and research applications that can most benefit from the new technology must be identified, and methods to take advantage of the combined data must be developed, according to Catana.
MORE FROM PHOTONICS MEDIA