Introduction to PET/MR Technology

Introduction to PET/MR Technology
PET/MR integrates two modalities that are highly complementary. Magnetic resonance imaging (MRI) has excellent soft tissue contrast and is itself a multi-modality imaging modality, where multiple types of image contrasts (T1, T2, DW, etc…) are attainable within one system. Positron emission tomography (PET) is a highly specific and sensitive molecular imaging modality that can identify small concentrations of radioactively-labeled tracers to probe function. Both modalities are capable of dynamic imaging

Integration of two modalities
Integrated PET/MR scanners have been available since 2011, even though the topic was first introduced 25 years ago. There is an exponential increase in the number of papers that describe MRI and PET each year, and this is expected to continue its growth. Integration of these two modalities requires that they exist both physically and electronically. Physical integration means that the PET and MR components must fit within the same space, essentially the PET scanner is placed within the MRI scanner itself. This requires the PET scanner crystals and detectors to be physically thin. It also requires careful consideration of all system power and cooling for both the MRI gradients and PET system which can also require active cooling. Electronic integration means that the PET and MR components must not interfer with each other such that electromagnetic interference due to RF and gradient pulses do not interfere with the PET system and vice versa. Vendors have largely solved these issues with the development of integrated PET/MR systems with modified body RF transmit/receive coils and novel PET detector systems. Most integrated PET/MR systems are built around a 70cm MR scanner where the resulting combined system has a 60cm bore. While this enables integrated simultaneous imaging, the resulting gradient performance is akin to a 70cm scanner with the patient limitations of a 60cm scanner.

Value of time of flight PET
Time of flight PET is technology available in many PET/CT and PET/MR scanners. This technique utilizes ultrafast PET detectors to measure slight coincidence timing differences between opposite detectors for a given pair of PET annihilation photons. It is particularly important because it provides improved image noise properties and improved sensitivity, particularly for large objects. More importantly for PET/MR, it reduces the bias in attenuation correction.

Attenuation Correction
Brief Overview
In PET, photons traveling through tissue undergo Compton scattering, for which when a photon is scattered it is no longer able to be detected. Thus the true number of photons measured is attenuated, hence the name attenuation correction. The likelihood of a photon undergoing Compton scattering is related to the linear attenuation coefficient, or essentially the density of the tissue or material. Therefore, in the human body, bone is much more likely to induce attenuation than soft tissue or lung tissue. While it is still possible to generate images without correcting for attenuation, there are noticeable and objectionable artifacts without the correction. With MRI, it is difficult to image bone with positive contrast, making PET/MR approaches to attenuation correction biased. Most MR-based attenuation correction methods utilize two-point Dixon chemical shift encoded imaging in a complicated image processing pipeline to yield a synthetic CT image that can be used to do attenuation correction. In many regions of the body, bone is ignored. As a result, these methods have known bias of greater than 15% for PET values in bone regions. However in soft tissue regions (including tumors within bone), the bias is on the order of 5% or less. It is unlikely that these biases limit the efficacy of PET/MR, most likely due to the increased diagnostic confidence given by the simultaneous MR imaging. New approaches such as whole body atlases to provide bone information reduce this bias. Recently, machine learning and deep learning have been applied to synthesize CT images with impressive results. It is likely that future MR-based attenuation correction methods will extensively leverage the power of these new and developing deep learning approaches.

Effect of MRI Contrast
The addition of contrast in CT imaging is known to affect quantitation from PET/CT images. Therefore, it is important to understand how the use of MR-based contrast agents may affect PET/MR. In general, at the higher energy levels of PET and the concentrations used in vivo, this is a minimal concern, and the use of MR contrast agents is not likely to affect the PET signal. However, MR-based attenuation correction methods utilize complicated image processing pipelines. These pipelines may be challenged by the use of contrast-enhanced MR images and potentially yield unexpected results in rare situations. While this is also a minimal concern, it is a reminder that the attenuation correction image should be visually inspected to ensure that the applied attenuation correction does not have any obvious artifacts.

Opportunities for motion robustness
A notable advantage of PET/MR versus PET/CT, is that MRI can be performed dynamically without a risk of increasing ionizing radiation dose to the patient. This allows for novel 4D attenuation correction schemes, which can reduce bias due to physiological motion.

Optimization of Protocols
PET/MR will benefit from the continued optimization of simultaneous protocols to take advantage of both modalities. In whole body PET/MR imaging, often the PET aspect can be performed faster per station than meaningful MRI series can be. For example when each body station is 3 minutes long, it is difficult to obtain even a T1 or T2 series in that timeframe in addition to the MR-based attenuation correction scan. However, in body region dedicated PET/MR studies, PET/MR is very efficient as the PET can be acquired over an extended period of time. Novel strategies to further optimize the time efficiency of PET/MR should be explored.

Acknowledgements

No acknowledgement found.