| Brain Imaging Centre |
| Montreal Neurological Institute (MNI) |
| McGill
University, Montreal, Qc. Canada |
| Research Projects in PET Instrumentation |
I have been involved in many important developments in PET instrumentation some of which are listed below. A bibliography according the different themes can be found by following the links below. In some cases copies of the articles listed in these pages are available. In june 2007 I retired from McGill and I am now a Professor Emeritus, still with close ties to friends at the MNI where I worked for 37 years.
Monte Carlo Simulation of PET Scanners. During the development of novel PET scanners it is important to estimate important parameters of the scanner's performance before committing to a specific design. The interactions of gamma rays with different materials can be simulated with Monte Carlo techniques when the interaction properties of the materials is known, and their size location and shape specified. Shortly after the Positome II (the first PET scanner to use bismuth germanate scintillation crystals as detectors) scanner was built at the Montreal Neurological Institute in 1978, the design was licensed to Atomic Energy of Canada Ltd. As part of their research program to optimize a commercial variant of this scanner, they developed a Monte Carlo simulation program specifically for PET scanners. Later we acquired the code and rights to this software and I adapted it to run of a VAX computer under the VMS operating system. The original code was extended my me and many students, and became known as "PETSIM". This software has been useful not only in the design of new PET scanners, but also in the understanding of performance of commercial PET scanners. It was used extensively in the design of the PEM-I scanner for detecting breast cancer, and for the optimization of removable shielding, the NeuroShield, for use during brain scanning on a conventional whole-body PET scanner. Thirteen peer reviewed papers have used this software or described its enhancements, and many more conference papers have discussed results obtained with it.
PET Detectors: their Structure, Limitations and Performance. The overall performance of a PET scanner is mainly determined by its detectors. Their size strongly influences the spatial resolution, their scintillation decay time and size influences the count-rate limitations of the scanner, and their encoding schemes influence the cost and spatial resolution. Over the years PET detectors have evolved from single crystals coupled to individual photo-multipliers (PMTs), to cut-block and clusters of optically coupled crystals coupled to four PMTs, or to position sensitive PMTs or avalanche photo-diodes (APDs). Some of the work in this area which was done in my lab includes the use of dual layer crystal block in which the crystals are offset by 1/2 of their spacing, improved encoding methods, measurement of the blurring due to coupling schemes and under-sampling.
Evaluation of, and improvements to, commercial PET scanners. Over the years the Montreal Neurological Institute has acquired several commercial PET scanners. Some of these were early models and their performance was evaluated here. One of these was the Scanditronix PC2048B, an early 15 slice brain scanner. Another example is the CTI ECAT HR+ scanner which is now used for all our human PET studies. During PET scans, most subjects move slightly, and some move far enough to make the results of questionable value. Monitoring subject movement with video cameras, and compensating for movement by encoding the subject's new position was first demonstrated on the Scanditronix scanner. A simple method for adding extra shielding to whole body PET scanners when they are used for brain imaging has been developed on the CTI HR+ scanner, and tested extensively. This concept, called the NeuroShield®, has been commercialized and is being manufactured by Scanwell Systems.
Timing properties of PET detectors. In order to reduce random counts which contribute to the noise in PET images, the timing window (during which two gamma rays are considered coincident must be as narrow as possible. However, if it is made too narrow, valid events are lost. Recently, much faster scintillation crystals like LSO have been introduced in commercial PET scanners, and these are gradually replacing the BGO crystals which were to norm in PET for over 20 years. Even though these are some seven times faster, the timing window in present "fast" PET scanners is typically only 1/2 of that used for BGO based scanners. One reason is the difficulty associated with measuring such small times, and another relates to the present techniques used to perform the time-alignment of PET scanners. Recently we have demonstrated the feasibility of performing the time alignment using a centrally placed positron emitting source embedded in a plastic scintillator. The plastic produces a substantial light flash as the isotope (typically Na-22) decays, just prior to the departure for the two 511 keV gamma rays which result from positron-electron annihilation. This serves as a reference time to which all the scanner's detectors can be time-aligned. Using such a source is expected to be very valuable for newer PET scanners which can encode the arrival time difference between the two gamma rays and use this time-of-flight information to improve the image quality. In order to use this device, and extra input for the reference clock is required, and this is not available on most scanners presently sold.
Positron Emission Mammography. This term and technique, which we proposed in 1994, to describe the use of small closely spaced PET detectors for use in the detection of breast cancer, has become the common name of this clinical imaging modality. During the next five years (with support for three years) from the National Cancer Institute of Canada's "Canadian Breast Cancer Research Initiative", we developed a dual-modality breast imaging concept by which the suspicious breast is first imaged by conventional X-ray mammography, and then with two PEM detectors which slide into place above and below the breast. Novel image reconstruction techniques were applied to allow for a real-time display of images form this instrument and we performed the first clinical trial of PEM on women with suspected breast cancer. This technique has now been commercialized and is sold by Naviscan PET systems.
Last updated:
October,
2009