Application of Annihilation Coincidence Detection to Transaxial Reconstruction Tomography (perspective on “Application of Annihilation Coincidence Detection to Transaxial Reconstruction Tomography” J Nucl Med. 1975;16:210–224)

The seminal paper by Phelps, Hoffman, Mullani, and Ter-Pogossian (1) from Washington University, St Louis, appeared just a couple of years after Godfrey Hounsfield presented the first CT images using x-rays. The CT scanner was a breakthrough that demonstrated the feasibility of imaging cross-sections of the human body without interference from the sections above and below. Even before Hounsfield’s pioneering CT device, David Kuhl and coworkers were exploring such a tomographic approach using single-photon counting in nuclear medicine with their Mark I–IV scintillation scanners, whereas Thomas Budinger and others used an Anger scintillation camera to acquire transaxial tomographic reconstructions. The first tomographic reconstructions using annihilation coincidence detection were obtained by David Chesler from Gordon Brownell’s group at Massachusetts General Hospital using a planar positron camera designed by the Boston group. Chesler applied filtered backprojection for the first time to medical imaging in 1971.

Phelps and coworkers were the first to propose a positron emission transaxial tomograph (PETT) comprising an array of individual detectors placed around a single transaxial plane. The design acquired transaxial tomograms of the positron-emitting distribution with optimized sensitivity for that plane. Furthermore, the reconstructed images exhibited a high signal-to-noise ratio and were quantitative in that the voxel concentration of a positron emitter could actually be measured. Earlier devices suffered from high dead times and recorded elevated rates of random and scattered coincidences, thereby reducing signal-to-noise ratio. However, the St. Louis design with individual detectors achieved depth-independent spatial resolution and sensitivity, a low dead time, and a low scatter fraction by applying an energy window and axial shielding. The uptake of the positron-emitting nuclide could be quantified through attenuation correction and calibration. The substantially superior performance of the annihilation coincidence detection system was further demonstrated by comparison with single-photon counting, where depth-dependent resolution and sensitivity, a lack of attenuation correction, and low intrinsic count rates resulted in high-noise, low-resolution, nonquantitative images. This comparison highlighted the difference between the “electronic” collimation capability of annihilation coincidence detection and the need for physical collimation with single-photon counting. The feasibility and performance of the PETT design were demonstrated by imaging phantoms and a dog. The results encouraged the group to develop a larger, single-plane PETT scanner that could accommodate a human (2), demonstrating for the first time that the concentration of a positron-emitting radionuclide could be measured quantitatively and noninvasively in human tissue (3).

The impact on PETT (subsequently termed PET, positron emission tomography) of this work from St. Louis in 1975 was profound, even though at the time very few centers worldwide were exploring the use of positron-emitting nuclides for medical imaging. Nevertheless, a sustained interest in PET led Phelps and Hoffman (after they relocated from St. Louis to UCLA) to collaborate with EG&G Ortec to design and build a commercial version of their prototype (1) that became known as the ECAT II (4). Several centers worldwide invested in this device, attracted by the ability to actually measure physiologic parameters such as blood flow, oxygen utilization, glucose consumption (5), and amino acid metabolism in both health and disease. Given this paradigm shift in medical imaging, interest in PET increased, and in 1981, a refined hexagonal array for brain imaging, the NeuroECAT, was developed, again in collaboration with EG&G Ortec. Then, in 1984, the hexagonal design of the 1975 prototype was replaced with a circular array of detectors in the ECAT III PET scanner. This circular geometry would become the design standard for future PET scanners.

A major limitation of the designs described above was that each detector comprised a scintillator coupled to a photodetector, an expensive and limited solution as the demand grew for higher spatial resolution and coverage of an increased axial length of the human body by incorporating additional rings of detectors. Around 1984, Computer Technology and Imaging (CTI), a spin-off from EG&G Ortec, developed a breakthrough block detector module (6) that allowed up to 64 scintillators to be coupled to only 4 photodetectors. The block detector provided a cost-effective solution for a PET scanner design comprising multiple rings of detectors. Compared with the pioneering PETT prototype, subsequent scanner generations have incorporated better spatial resolution, greater sensitivity, faster data acquisition, lower random and scattered coincidence rates, lower detector dead time, greatly increased axial coverage, and recording of time-of-flight data. The fundamental concept of electronic collimation has been extended to fully 3-dimensional acquisition, eliminating the need for interplane shielding, and, more recently, the high-resolution molecular images have been complemented with high-resolution anatomic images through the combinations of PET/CT (in 2001) and PET/MR (in 2010).

The 24-detector PETT was able to record, within a 2.5-cm axial and approximately 16-cm transaxial field of view, only 12 coincidence lines of response at any one time, and with rotation and linear offset a total of 1,728 lines of response could be recorded for the reconstruction. At the conclusion of their paper, Phelps et al. speculated on the potential to increase the number of lines of response by extending the electronic collimation concept. Nothing illustrates the tremendous progress in PET scanner design over the past four and a half decades more than the fact that the current EXPLORER total-body PET scanner incorporates 564,480 detectors and 92 × 10⁹ lines of response and covers 198 cm axially of the human body.

• © 2020 by the Society of Nuclear Medicine and Molecular Imaging.

• Received for publication May 4, 2020.
• Accepted for publication May 7, 2020.