Integration of PET with Magnetic Resonance Imaging (MRI) without Mutual Interference
PET is very sensitive at detecting and quantifying a low abundance molecular target on or within diseased cells. MRI is capable of providing exquisite anatomy as well as physiological measurements of disease states. Thus, the two modalities provide complimentary information, and as a result, recently there have been efforts to combine the two modalities by inserting a PET system into an MRI system and operating them simultaneously for highly accurate spatial and temporal registration. If successful, the result would be a powerful tool to characterize disease in living subjects. However, this is a very difficult instrumentation problem since with existing PET and MR technologies, there would be substantial mutual interference between the two modalities: The standard PET system design would have a significant electrical footprint, which would interfere with the MRI system’s performance, and the latter creates an extremely noisy environment for the former to operate. Siemens has built a prototype brain-only PET insert for their MR system, by adding conductive shielding around the PET components; but scaling that design up to a whole-body version, which is currently the clinical “standard-of-care”, would mean a log-order increase in the electrical footprint (e.g. the volume of conductors and coaxial shielded cables used for signal transmission out of the MRI system would greatly increase).
We are addressing this considerable engineering problem by drawing from the field of optical communications. It turns out that the analog PET scintillation detector signals can directly couple to and be used to drive a telecommunications-grade vertical cavity surface emitting laser (VCSEL), and all the required information may be carried out as light pulses using multi-mode fiber optics, rather than as electrical signals down shielded coaxial cables. As long as the front-end scintillation detector components (scintillation crystals and photodetectors) as well as the VSCEL can be made non-magnetic (which we have shown), and introducing this electro-optical coupling step does not introduce temporal or amplitude dispersion (which we have shown), the result would be a system with very low electrical footprint (no signal cables or shielding required), scalable to a whole body PET geometry.