At the time of Meneely and Hahn were working in 1943, the only radiation detectors in general use were gas-filled Geiger-Muller counters, and ionization chambers. Radioisotope measurements were accomplished using these non imaging devices for sample counting in order to estimate absorption, fate and excretion of administered radioactivity before the availability of imaging devices. The importance of static imaging was recognized as pioneering investigators moved small shielded hand-held GM tubes step-wise over the neck of patients to record the distribution of functional activity on a grid displaying I-131 counts sensed over different regions of the thyroid gland.
In 1949, Ben Cassen working in Stafford Warren’s University of California, Los Angeles AEC laboratory developed the first radioisotope scanner for medical imaging. Although NaI had been discovered several years earlier by Hofstader, Cassen attached a cadmium tungstate(CdW04) scintillator to a photomultiplier tube attached to a motor driven device that moved the collimated probe in a rectilinear raster over the subject. CdW04 had an advantage over NaI-based scintillators based on its large photo fraction. Warren was a radiobiologist trained at the University of Rochester who had held the top health and safety position in the Manhattan Project, and was involved in the Bikini weapons tests, and the earliest Japanese A-bomb survivor studies. Like others in key positions in the A-bomb project, he was a strong proponent for nuclear medicine, and helped push the program at local (UCLA) and national levels. Larry Curtiss founded a company that marketed the Curtiss-Wright clinical scanner for a brief time before Picker developed and marketed a more advanced device using NaI (Tl) as the scintillator in the late 1950s. About the same time, Val Mayneord in the UK also was developing radionuclide imaging devices.
In addition to single photon gamma ray scanning devices, several positron scanning devices were developed and successfully used in the mid 1950s for imaging PET tracers in patient and animal studies. The MGH/MIT group led by Brownell developed a scanner with dual opposed probes that were used to image positron emitting tracers in brain tumor patients[[iv]]. Similar devices were developed in Canada and Europe, and innovations rapidly progressed on many fronts. Indeed, the prospects for successful tumor imaging with PET tracers looked very bright from the beginning when high contrast images were demonstrated in patients with brain tumors using radioactive arsenic-labeled compounds.
Hal Anger came to the University of California in Berkeley (UCB) in 1946 from Harvard (the Rad Lab), to the Donner Lab where he spent the rest of his career until he retired in 1982[[v]]. The first of his many nuclear-related inventions was the NaI well-crystal detector. The well-type crystal provided increased sensitivity by increasing the geometric coverage of samples contained in test tubes where the sample was inserted into and surrounded by NaI on all sides. He went on to develop a series of instruments and procedures in support of many imaging applications. A first task was to provide engineering support for the Berkeley cyclotron-based alpha particle pituitary radiation therapy program. In 1952, Anger and Bob Mortimer, a Medical Physicist in the Donner laboratory, developed a gamma camera using a Westinghouse image intensifier tube that Russell Morgan had invented for x-ray imaging. The gamma imaging device proved to be too insensitive for clinical imaging and Anger went on and developed the NaI crystal/PMT camera design that continues to be the currently used clinical gamma ray imaging device. Around the same time as Anger was developing the gamma camera, in 1956 he designed and built the first multi detector whole body scanner.
Meneely at Vanderbilt collaborated with the Physics Department (Prof. Sherwood Haynes and two Physics graduate students: Robert Kerr, and Jesse Hoffman) in the design of a gamma-ray imaging camera. The device used a small, 1″ diameter rotating oscillating pin hole collimator backed by a 1″ thick NaI crystal. The collimator allowed radiation to reach the detector when pointed at the direction of the in-coming collimator-defined beam. Such events were recorded on film by coupling the collimator motion to that of an oscilloscope readout device. The camera was used to record the distribution and intensity of 412 keV gamma rays emitted from Au-198 colloid injected into a patient. The work was presented at an American Physics Society meeting that took place in Knoxville, TN April 1-3, 1954 with the title “A Rotating Sphere Solid Angle Scanner for Gamma Rays”. The only other published record of the work was a one page illustration in Nucleonics in 1955 entitled Spinning-Ball CRT Scanner with an image of a patient given Au-198 colloid[[vii]]. A similar optical readout coupling method was used by Hal Anger for his first gamma camera readout, and by Paul Harper for his first whole body image made with a longitudinally scanning small gamma camera with a similar motion-coupled oscilloscope readout. The inherent problem that limited Meneely’s imaging device was its very low sensitivity due to the small solid angle subtended by the collimator, and its small low sensitivity NaI scintillator. The image was acquired in 5 minutes, but although unstated in the article, the poor quality image was acquired from a patient that received a therapy dose. The device was not further evaluated as it was not suited for use in patients for clinical studies.
Anger by 1956 had a working pin-hole collimated camera in which 7 photomultiplier tubes were close packed on a 4” diameter NaI crystal. Data from the multiple phototubes were used to read out the signals from the light flashes from energy stopped in a single small diameter crystal. The x-, y- address and z- energy of each event was decoded by an X-Y-Z summation circuit that became known as the Anger logic. Bob Mortimer went on and had a distinguished career advancing fundamental understanding of genetic effects of radiation. Based on extensive studies in yeast he was awarded the prestigious Beadle prize in Genetics. He served as Dean in the Graduate School, and was interim Director of the LBL Genome Project before retiring.
Images obtained with Anger’s camera imaging device were first presented at the United Nations Conference on the Peaceful Atom, and significant interest was elicited for clinical uses. Bill Myers, a nuclear medicine physician/scientist on the faculty of Ohio State University visited Berkeley frequently, working with Anger on the prototype device on several projects. He was so impressed that he convinced John Kuranz, the President of Nuclear Chicago, to build the device for commercial use, and was the first to acquire one in 1962 for his use at Ohio State. Nuclear Chicago morphed from a small company known for its multichannel analyzer product to the premier manufacturer of Anger-type Gamma Cameras. The product was an instant hit as it gave practitioners an effective easy-to-use, multi purpose imaging device, which to this day is the main instrument used by nuclear medicine practitioners around the world.
An important alternative to the Anger camera was the Autofluoroscope developed by Merrill Bender and Monte Blau. In Craig Harris’s presentation of their Hevesy award, he noted that Its high data rate capability made it possible to image fast dynamic processes including first pass studies. Being a stationary device also made it possible to correct for motion distortions, not possible with scanning devices.
Probably the most important scanner innovation was made by Kuhl and Edwards at the University of Pennsylvania. Between 1960 and 1964, they developed the first practical longitudinal and transverse section scanning tomographic device which paved the way for the subsequent development of X-ray CT. They developed their own electronics, and built a structurally formidable dual headed scanner that just fit into a standard sized imaging room. Image formation used simple first order back projection methods to display the images which were better than any that had been collected prior thereto. Their work no doubt laid the ground work for Hounsfield and Cormack who went on to get the Nobel Prize for X-Ray tomography using more sophisticated image reconstruction methods they adapted from prior work done by astronomers. Over the years, Kuhl went on to develop a series of specialized head-only imaging instruments using more stream-lined but similar approach. They produced much improved cross sectional brain images that allowed one to clearly differentiate brain tumors from stroke lesions.
It has always been difficult to convince industry that special purpose dedicated devices would be profitable. The first computer-controlled dual-head scanner, which produced longitudinal and transverse section images of gamma rays, was developed by Kuhl and Edwards at the University of Pennsylvania. Subsequent evolutions improved system sensitivity and resolution by increasing the number of detectors, and a more-efficient scanning geometry. The device produced high-quality transverse-section images of the brain which distinguished tumor masses from wedge-shaped lesions characteristic of stroke. It became a brain-only imaging system, but none of the companies at that time were willing to produce and market a device for s a limited market. As time proceeded, they were proved wrong and specialized brain and heart imaging devices are used in many imaging modalities. The commercial success of X-ray computed tomography changed the instrumentation picture enormously. It showed that a device that was clinically useful and accomplished important tasks could be sold despite high cost. This lesson carried over to Magnetic Resonance Imaging (MRI), and the escalating complexity of modern gamma-ray imaging devices, including PET, PET/SPECT, and CT coupled nuclear imaging devices. It turns out that a device will sell if it can do something worthwhile that can not be accomplished otherwise.
The development by Beck and the Argonne Cancer Research Hospital group of a 4 crystal scanner (ACRH Scanner), optimized for imaging 99mTc, ushered in a new generation of devices taking advantage of the newly emerging applications of the 140 keV tracer which was responsible for the rapid growth spurt in clinical nuclear medicine.
At Vanderbilt, in 1969/70, we obtained one of the early dual opposed 8” diameter scanners from Ohio Nuclear. The electronics and display was developed by Jim Mozely at Johns Hopkins, and the system worked very well. We used the 8” diameter collimators for patient scanning, but investigated the effect of collimator solid angle on lesion detection by systematically narrowing the collimator aperture by filling outer holes with powdered lead. We imaged known objects with known activity at different depths in water phantoms, synthesized virtual images by scaling and adding them to normal brain scans and compared detection thresholds and accuracy with collimator aperture size. Structured background added to the complexity of the detection task, as did the noise structure as both lesion size and activity content diminished. The large 8” crystal did not deliver as much benefit as had been anticipated based on its higher total sensitivity, and it awaited the development of tomographic imaging systems to over come this limitation.
The most commonly used gamma-emitting tracer used in the early days was reactor-produced 131I, which also emits a number of energetic beta rays. To keep radiation doses acceptably low, only small amounts can be administered safely or diagnostic studies especially when using long-lived beta/gamma emitting tracers. To efficiently stop and collect the energy from 131I requires thick detectors and thick 2”, 3”, 5”, and 8” diameter NaI (Tl) crystals came into standard use in later years. The collimators used with the larger area detectors produced hour glass shaped response profiles with best resolution at the focal point of the system. The smaller diameter collimated detectors had response profiles shaped like elongated ellipses and these provided the more uniform depth response needed for cross sectional tomographic imaging that was then emerging.
A major innovation in scanners was made by Anger when he developed the Multiplane (longitudinal) tomographic scanner. The device was a camera-based rectilinear scanner, marketed as the Pho-Con, and its thick NaI crystal made it well suited for 131I and 67Ga imaging for which the device is still in clinical use primarily in Europe. The gamma camera makes better 99mTc images and the Pho-Con was less well accepted and used in the USA.
Anger’s Mark II whole body scanner consisted of an array of 64 NaI detectors with a single bank of focused collimators below the subject. The patient was moved over the detector array and the whole body of the patient was imaged in a matter of several minutes. Ostertag (DKFZ, Heidelberg) subsequently extended this approach to whole body positron imaging by the addition of an upper bank of detectors which operated in coincidence with the opposed crystal array. The efficiency of positron and gamma-ray imaging systems depends basically on the amount of scintillator and the geometry used in the imaging system. Stopping the 511 keV annihilation photons requires thicker crystals than are needed for low energy gamma imaging. The number, kind, and volume of the detectors, their geometric arrangement and the criteria for accepting a valid event, influence the relative sensitivity of the two methods. Using coincidence electronics, the DKFZ device produced whole-body emission (and transmission) images of annihilation photons with resolution appropriate for their sequential dosimetry studies.
Many other approaches to obtaining longitudinal tomography were also pursued. George Taplin developed a large diameter hemispherical detector which focused at a selected depth. It produced good images at the focal plane but blurred images from other depths. Muehllehner invented a rotational slant hole collimator which allowed one to obtain dynamic information at multiple depths. It depended on the reconstruction algorithm to separate activity from over lapping planes using a stationary detector. The image reconstruction logic was similar to the Anger PhoCon approach, but used a stationary detector unlike the rectilinear scaner approach used with the PhoCon. The problem with the rotational collimator was that there were artefacts due to the use of relatively primative reconstruction methods. At Vanderbilt, Jon Erickson coupled the PhoCon, as the Multiplane Anger camera came to be known, to a PDP/9 and used the computer to display user-defined images at multiple levels.
Many different gamma camera embodiments and applications have since been realized. Anger and Muehllehner rotated a patient in front of the camera and made tomographic images analogous to those produced by Kuhl. Harper and Beck translated a person through the small field of view of the gamma camera and moved the optics of a readout display to accomplish the first gamma camera whole body scan. Ron Jaszczak and John Keyes adapted rotating gamma cameras for transverse section imaging. Muehllehner developed rotating slant hole collimators which produced longitudinal tomographic images with some artifacts that required more sophisticated image processing than had yet emerged. Many of the early camera developments were made by Nuclear Chicago, its successor organizations, and a cadre of their spun-off scientists (Jaszczak/Muehllehner etc.). Subsequently, GE, Picker, Siemens, Philips, El Scint, Toshiba, Hitachi, and a few other companies produced commercial gamma camera based systems.
A great deal of work was done to optimize collimators for scanning and camera-based imaging systems. Camera systems typically employ parallel-hole collimators, where hole size, septal thickness and collimator length are designed based on a trade-off between the desired spatial resolution (at appropriate distances), and sensitivity for the energy range that will be used. Similar decisions are needed for the design of slant hole, converging, multi and single pinhole collimators[xiii]. Collimator design has improved based on analytical insights, and on the use of Monte Carlo simulation methods.
Positron-emitting radionuclides are mostly cyclotron-produced and hence early developments with short-lived emitters were limited to places with on-site cyclotrons. Longer-lived positron-emitters and generator-produced positron emitters were used at more remote sites. The scanner mode of operation was further limited to imaging static, or very slow dynamic, processes. By the end of the 1950s, it was appreciated that mapping the distribution of positron-emitting tracers could have a significant role for detecting and localizing brain tumors in patients. The advantage over single-photon imaging was due to the better tumor-to-background ratio associated with the higher tumor avidity of the positron-emitting radiopharmaceuticals that were then available. Much of the early work used tracers with several-day half-lives, mainly for brain-tumor imaging. Initial studies produced high-contrast diagnostic images using As-74 (17.5 d) and Cu-64 (12.8 h) tracers, and enthusiasm for their development and use was high. Stationary pairs of opposed non imaging counting probes were used by the Hammersmith group to map regional transients of respired gases. Kinetics of 15O, 13N, C15O2 and 11CO2 gases were measured in the upper, middle and lower thirds of the human lung in the field of view of the paired detector sets. Much of the modern pulmonary physiology knowledge of pulmonary blood flow and gas exchange was derived from these pioneering studies.
Major advances have been made in PET imaging systems since the early dual opposed scanning devices. The first of the important devices was the hemispherical detector developed by Robertson at BNL in 1960. Mathematical tools for reconstructing the data were investigated by Bob Marr but clinical application awaited Yamamoto and Thompson at the Montreal Neurological Institute who collapsed the design to a single ring cylindrical array and used the modified device in path finding research. Ter-Pogosian, Hoffman, Mullani and Phelps at Washington University worked with their Biomedical Computing Lab and developed a series of cylindrical ring systems which are the typical configuration used clinically to this time. Brownell at MGH/MIT developed a series of important PET imaging devices. The first was a large area dual opposed planar array. Its large area made it uniquely suited for pulmonary imaging. With the exception of illustrative studies in the 1970s by the MGH group (Brownell and Hoop) little 3-or 4-D imaging of lung function was performed to expand on the earlier probe based studies by West et al in the UK. The lack of a large-field-of-view commercial ring based positron-imaging device precluded the further development of pulmonary imaging for many years, and even today little work is being done in this important area.
Brownell developed advanced coding schemes to reduce the component cost; much as was used by Bender and Blau on their Autofluoroscope, that became the Baird Atomic planar fast high energy gamma imaging system. Brownell and Burnham also developed a cylindrical array system as did Z H Cho, using coded detector arrays. The most recent systems have used block detectors that permit imaging multi z axis planes with good x-, y- position within planes. Clever packaging of BGO, NaI, LSO, YAP, GSO scintillator blocks are being used in modern systems trying to use new crystal and electronic tools to best advantage. In order to over come distortion particularly at the outer regions of the body, clever means of encoding depth of interaction have been developed. These range from the use of opposed arrays of APDs to decoding wave form shapes from different elements in multi level (4x4x4x4) bocks, and advanced form of Phoswich detectors. Along the way, Allemand and the Grenoble group developed a time of flight (TOF) system using BaF2, because of its fast scintillation properties. Newer LaBr2 crystals show promise and are being used for TOF. Bill Dunn, a graduate student at Vanderbilt in the early 1970s tried using NE-111, a fast low efficiency plastic scintillator coupled to a pair of RCA experimental phototubes. The 250 ps uncertainty he achieved was felt to be too long and the number of detector pairs needed to populate ring geometry was thought to be too large to proceed further with that approach. Since that time, faster scintillators have become available, and TOF PET has be come a clinical reality using LSO ring detectors. The advantage turns out to be primarily in very obese patients, where the fractional uncertainty in positioning the event is a small fraction of the patient’ girth.
The Michigan group led by Rogeers and Clinthorne has done much of the modern work that has advanced the Compton Coincidence device pioneered by Manbir Singh in the early 1980s. Promising improvements in sensitivity and resolution are under study using a silicon detector ring inserted into a commercial PET ring, and both 511 keV singles, coincidences, and CCI interactions are recorded and used in image reconstruction[[xviii]]. Compton imaging continues as an elusive but attractive goal, especially for high energy imaging.
Improved collimator designs continue to evolve. A fundamental advance was made by Barrett and the Arizona group when they developed the first fast 3D dynamic brain and then heart imaging device. This used multiple detectors in a stationary ring about the patient. Each of the detectors collected energy and address information, and advanced signal processing was used to calibrate the system, and facilitate optimal reconstruction using computationally enhanced data capture and reconstruction approaches. They extended this to higher resolution when they changed from compact gamma cameras to a large number of ultra high resolution pin hole collimated small detectors, each of which had a very high space bandwidth product, i.e. a large ratio of resolvable elements per unit surface area. By increasing the number of detectors they overcame the low sensitivity of each element in the system. The best devices used CZT but materials quality problems slowed down the work. The concept can be used for human, and large animal systems, but the initial success was with small animals, mice and rats due to detector materials and size issues. While this work was progressing, Beekman in Utrecht developed a series of spherical collimators with ultra small pin holes focusing toward the center of the sphere, where a small animal could be situated. With 2 and 3- headed commercial camera he demonstrated the feasibility of 200-500 μm resolution in small regions, such as the heart or brain of a mouse.
Commercial manufacturers picked up ideas from University-based research innovations and over the ensuing years added to them as they produced a number of multi-detector devices using arrays of focused detectors moved above and below patients to produce maps of organ uptake and whole-body distribution images. Some of the devices had multiple scanning detectors arrayed at different angles, each of which moved in different patterns about the patient. The major aim of these developments was to increase system sensitivity (by the adding imaging receptor area) and resolution (by improved spatial sampling) all of which can lead to improved image quality.
The use of digital electronics was able materially improve gamma camera count rate performance when one became able to sample the output of the PMTs, and process archived list mode data. has led to numerous improvements in gamma camera performance. It makes it possible to use the same device for SPECT and PET as one needs to change the PMT pick off point and electronic gating differently for each application, something that Muehllehner was unable to do in his late 1970s device. Signal processing has now made it possible to increase the count rate performance of gamma cameras to 2-4 MHz, instead of the 50-200 KHz of previous devices, although this is not yet incorporated into commercial devices.
Solid state imaging devices have been used in research applications since the 1970s, but have been slow to appear in commercial devices. Digirad attempted to introduce a cadmium zinc telluride (CZT) imaging receptor for cardiac applications in 1998. Difficulties in assembling large arrays impeded the commercial success and they switched to CsI. Several manufacturers in 2006 used CZT in tomographic devices. D-SPECT introduced a novel approach which uses an ensemble of 8 tungsten collimated CZT detectors which are directed toward the heart. Each swings through an arc which covers only the cardiac emitting region. Cardiarc uses stationary detectors, but a rotating slit slot collimator that sorts data from the different emitting regions. The increased sensitivity and preserved resolution provides better images in shorter times.
Orthogonal strip Si(Li) and Ge(Li) detectors have been used for research studies for low energy and high energy imaging respectively. The former have high potential for imaging of I-125 labeled compounds especially in animal studies. Compton Cameras have proven their value in astronomy by imaging planar distribution of high energy gamma rays from distant stars. Compton Cameras face a more complex near field 3D imaging task in medical applications and despite much effort by different talented researchers, a practical clinical device is still awaited. Recent technological advances in scintillator materials, solid state detectors, photomultipliers, avalanche photodiodes, and computer-based data acquisition, processing and display systems have accelerated progress since the initial work by Singh[[xxiii]]. New success is based on spin-offs from NASA, DOD and DOE research[[xxiv]]. Other advances are being realized from Astronomy developments using high resolution time projection chambers with Anger Camera absorption detectors for Compton Imaging[[xxv]]. Significant improvements in image reconstruction methods have been achieved extending List Mode MLEM algorithms developed by Barrett[[xxvi]] and the Michigan group[[xxvii]]. Walenta and Brill worked together on the development of a Compton Camera since 1980 when they were both at Brookhaven National Lab, and have a continuing collaboration thereafter[[xxviii]]. An NIH STTR grant received Phase 1 funding for a project entitled “High Energy Imaging in Support of Radionuclide Therapy”. The project failed to get Phase 2 support due to short comings in the Commercialization Plan, but was directed at support for a 212Bi alpha particle based antibody targeted therapy of malignant melanoma where the task was to image the emitted 2.6 MeV gamma ray in patient studies. The VU collaborators included Todd Peterson, Mike Stabin, and Randy.
→ Instrumentation History: At the time of Meneely and Hahn arrived at Vanderbilt in 1943, the only radiation detectors available for clinical and research measurements were gas filled Geiger-Muller counters, and ionization chambers. Radioisotope measurements were accomplished using these non imaging devices for sample counting to measure absorption, fate and excretion of administered radioactivity in the absence of imaging devices. Imaging was first conducted by dedicated investigators who moved small shielded hand held GM devices step-wise over the neck of patients to record counts on a grid from I-131 emanations from different regions of the thyroid gland. Click here for a web history of nuclear instrumentation.