Endogenous Technology Adoption and Medical Costs.

Despite the claim that technology has been one of the most important drivers of healthcare spending growth over the past decades, technology variables are rarely introduced explicitly in cost equations. Furthermore, technology is often considered exogenous. Using 1996-2007 panel data on Swiss geographical areas, we assessed the impact of technology availability on per capita healthcare spending covered by basic health insurance whilst controlling for the endogeneity of health technology availability variables. Our results suggest that medical research, patent intensity and the density of employees working in the medical device industry are influential factors for the adoption of technology and can be used as instruments for technology availability variables in the cost equation. These results are similar to previous findings: CT and PET scanner adoption is associated with increased healthcare spending, whilst increased availability of percutaneous transluminal coronary angioplasty facilities is associated with reductions in per capita spending. However, our results suggest that the magnitude of these relationships is much greater in absolute value than that suggested by previous studies that did not control for the possible endogeneity of the availability of technologies. Copyright © 2016 John Wiley & Sons, Ltd.

3D scintigraphic imaging and navigation in radioguided surgery: freehand SPECT technology and its clinical applications.

Freehand SPECT (fhSPECT) is a technology platform for providing 3-dimensional (3D) navigation for radioguided surgical procedures, such as sentinel lymph node (SLN) biopsy (SLNB). In addition to the information provided by conventional handheld gamma detection probes, fhSPECT allows for direct visualization of the distribution of radioactivity in any given region of interest, allowing for improved navigation to radioactive target lesions and providing accurate lesion depth measurements. Herein, we will review the currently available clinical data on the use of fhSPECT: (i) for SLNB of various malignancies, including difficult-to-detect SLNs, and (ii) for radioguided localization of solid tumors. Moreover, the combination of fhSPECT with other technologies (e.g., small field-of-view gamma cameras, and diagnostic ultrasound) is discussed. These technical advances have the potential to greatly expand the clinical application of radioguided surgery in the future.

Medical physicists' implication in radiological diagnostic procedures: results after 1 y of experience.

Since January 2008-de facto 2012-medical physics experts (MPEs) are, by law, to be involved in the optimisation process of radiological diagnostic procedures in Switzerland. Computed tomography, fluoroscopy and nuclear medicine imaging units have been assessed for patient exposure and image quality. Large spreads in clinical practice have been observed. For example, the number of scans per abdominal CT examination went from 1 to 9. Fluoroscopy units showed, for the same device settings, dose rate variations up to a factor of 3 to 7. Quantitative image quality for positron emission tomography (PET)/CT examinations varied significantly depending on the local image reconstruction algorithms. Future work will be focused on promoting team cooperation between MPEs, radiologists and radiographers and on implementing task-oriented objective image quality indicators.

Imaging dose in variable pitch body perfusion CT scans: an analysis using TG111 formalism.

PURPOSE: To investigate the variation of imaging dose with tube potential in variable pitch body CT perfusion (CTp) protocols using the TG111 dosimetric formalism. METHODS: TG111 recommendations were followed in choosing the phantom, dosimetric equipment, and methodology. Specifically, equilibrium doses (D(eq)) were measured centrally and peripherally in a long PMMA phantom. Reference planar average equilibrium doses were determined for each tube potential, for a reference set of exposure parameters (collimation, pitch, filtration) on a Siemens Definition CT scanner. These reference values were utilized to predict the imaging dose during perfusion scans using interpretations of the TG111 formalism. As a gold reference, the midscan average planar perfusion doses (D(CTp)) were obtained directly from central and peripheral D(eq) measurements for body CTp scans (144 and 271 mm) using variable pitch acquisition. Measurement-based D(CTp) values obtained using a thimble chamber were compared to the TG111-predicted values, and to CTDI(vol) reported at the console. RESULTS: Reference planar average equilibrium dose values measured for reference uniform pitch helical scans were consistently higher than console-reported or measured values for CTDI(vol). The measurement-based perfusion dose D(CTp) was predicted accurately by the reported CTDI(vol) for the 144 mm scan. The 271 mm scans delivered systematically larger dose than reported. The TG111-based dose estimates were proven to be conservative, as they were systematically higher than both the measured and the reported imaging doses. CONCLUSIONS: Upon successful implementation of TG111 formalism, standard imaging dose was measured for a body CTp protocol using the variable pitch helical acquisition. The TG111 formalism is not directly applicable to this type of acquisition. Measurement of dose for all variable pitch protocols is strongly suggested.

Low-activity (124)I-PET/low-dose CT versus (131)I probe measurements in pretherapy assessment of radioiodine uptake in benign thyroid diseases.

CONTEXT: Radioiodine therapy of benign thyroid diseases requires pretherapy assessment of radioactive iodine uptake (RAIU) for reliable therapy planning. OBJECTIVE: Our objective was to assess RAIU by low-activity (124)I-positron emission tomography/low-dose computed tomography ((124)I-PET/CT) in comparison with standard (131)I probe measurements. DESIGN/SETTING: This prospective comparative study was conducted at the Jena University Hospital, Jena, Germany, in a referral center setting. PATIENTS: A total of 79 patients with benign thyroid diseases were screened, 40 of whom met the inclusion criteria (stable TSH, free T3 and free T4 levels; no thyroid-specific medication, no iodine contamination) and 24 of whom agreed to participate by signing an informed consent. INTERVENTIONS: All patients received the standard (131)I scintillation probe uptake test 30 hours after administration of 3 MBq (131)I. Seven days later, all patients were subjected to (124)I-PET/CT uptake measurement 30 hours after administration of 1 MBq (124)I. MAIN OUTCOME MEASURES: The decay-corrected uptake values of both techniques were compared. Additionally, 3 different volume-of-interest-based evaluation methods in PET/CT (whole neck [WN], automatic isocontour [IC], and manually contoured [MC]) were evaluated. RESULTS: The (131)I probe measurement and (124)I-PET.WN method provided very similar mean RAIU (30.7% ± 10.3%; 31.7% ± 8.9%), resulting in a significant positive correlation (r = 0.93, P < .001). Compared with (124)I-PET.WN, the (124)I-PET.IC (29.8% ± 8.6%) and the (124)I-PET.MC (24.5% ± 7.1%) demonstrated lower uptake values. CONCLUSIONS: Using activities as low as 1 MBq, the (124)I-PET.WN method shows a good correlation with conventional (131)I probe measurement. Thus, (124)I-PET/CT is a suitable alternative for pretherapy RAIU evaluations. This may offer potential additional benefits such as PET/ultrasound fusion imaging and CT volumetry.

[ECONOMIC EFFICIENCY TECHNOLOGIES DIAGNOSIS OF ISCHEMIC HEART DISEASE BY MAHNITOTOKARDIOHRAFIYI].

The article is devoted to clinical--economic analysis of modern diagnostic technology--magnetocardiography by analyzing the "cost-effectiveness". Economic effectiveness of diagnosis of coronary artery disease using magnetocardiography in terms of cost/effectivness is shown. The economicaly optimal sequence of several noninvasive methods for diagnosis of coronary artery disease is defined.

An animal PET scanner using flat-panel position-sensitive PMTs.

OBJECTIVE: To design, build, and evaluate an animal PET scanner, which can be used with non-human primates under conscious condition, incorporating flat-panel position-sensitive photomultiplier tubes (PS-PMTs). METHODS: The system contains 30 detector modules, each having two PS-PMTs and 16×18 lutetium­yttrium oxyortho-silicate scintillation crystal arrays. The system has 17,280 crystals (480 per ring) arranged in 36 rings, with a diameter of 508 mm and axial extent of 108 mm. The gantry tilt mechanism enables PET studies to be performed on a monkey in the sitting position. Data can be acquired in either the 2D or 3D mode, with the slice collimators being retracted in the 3D mode. RESULTS: At the center of the field-of-view, radial resolution is 2.7 mm full width at half maximum (FWHM) and tangential resolution is 2.4 mm FWHM, while axial resolution is 2.5 mm FWHM for direct slices and 2.7 mm FWHM for cross slices. Scatter fraction, count rate capability, and sensitivity were evaluated using a cylindrical phantom 10 cm in diameter. The noise equivalent count rate in the 3D mode is equivalent to that in the 2D mode at a three times higher radioactivity level. Total system sensitivity is 1.3 kcps/(kBq/mL) in 2D mode and 7.4 kcps/(kBq/mL) in the 3D mode. Animal studies with a monkey were performed to evaluate the imaging capabilities of the scanner. CONCLUSION: The new PET scanner will be a useful research tool with non-human primates for pre-clinical drug development.

A macroquantification approach for region-of-interest assessment in emission tomography.

In this article, we propose a quantification methodology for estimating the statistical parameters of the activity inside regions of interest (ROIs). Macroquantification implies a rearrangement of the emission projection data into macroprojections and a redefinition of the system matrix based either on an image reconstruction involving iterative ROI-wise regularization or on an ROI uniformity assumption. The technique allows a very fast computation of the ROI activities and covariance matrix in the least squares sense using a low-dimensional model of the tomographic problem. The macroquantification approach is evaluated through Monte Carlo simulations using a numerical thorax phantom, without taking into account the measurement artifacts and assuming a perfect a priori ROI definition. Various tumor ROI configurations and count rates are considered to reflect clinical situations. The results show that our technique yields low-bias ROI estimations that turn out to be more accurate than classical estimates relying on pixel summation. Macroquantification also provides an approximation for the ROI variance that describes the effective variance obtained through the simulations fairly well. The technique is then validated using single photon emission computed tomography (SPECT) data from a physical phantom composed of cylinders filled with different Tc concentrations for the task of ROI comparison. Here again, the study shows excellent agreement between the measured and predicted values of the ROI variance resulting in efficient estimations of ROI ratios and highly accurate ROI comparisons. In its simplest formulation, macroquantification has a short computation time, making it an ideal technique for quantitative ROI assessment that is compatible with a wide range of routine clinical applications.

Accurate modeling of a DOI capable small animal PET scanner using GATE.

In this work we developed a Monte Carlo (MC) model of the Sedecal Argus pre-clinical PET scanner, using GATE (Geant4 Application for Tomographic Emission). This is a dual-ring scanner which features DOI compensation by means of two layers of detector crystals (LYSO and GSO). Geometry of detectors and sources, pulses readout and selection of coincidence events were modeled with GATE, while a separate code was developed in order to emulate the processing of digitized data (for example, customized time windows and data flow saturation), the final binning of the lines of response and to reproduce the data output format of the scanner's acquisition software. Validation of the model was performed by modeling several phantoms used in experimental measurements, in order to compare the results of the simulations. Spatial resolution, sensitivity, scatter fraction, count rates and NECR were tested. Moreover, the NEMA NU-4 phantom was modeled in order to check for the image quality yielded by the model. Noise, contrast of cold and hot regions and recovery coefficient were calculated and compared using images of the NEMA phantom acquired with our scanner. The energy spectrum of coincidence events due to the small amount of (176)Lu in LYSO crystals, which was suitably included in our model, was also compared with experimental measurements. Spatial resolution, sensitivity and scatter fraction showed an agreement within 7%. Comparison of the count rates curves resulted satisfactory, being the values within the uncertainties, in the range of activities practically used in research scans. Analysis of the NEMA phantom images also showed a good agreement between simulated and acquired data, within 9% for all the tested parameters. This work shows that basic MC modeling of this kind of system is possible using GATE as a base platform; extension through suitably written customized code allows for an adequate level of accuracy in the results. Our careful validation against experimental data confirms that the developed simulation setup is a useful tool for a wide range of research applications.

Development of a new, fast, user friendly, ray tracing program "CSIM" for the simulation of parallelhole collimators.

We have developed a fast, user friendly, ray-tracing program, "CSIM" for low-energy gamma rays (up to ∼200keV) to simulate the performance characteristics of parallelhole collimators. We have used a ray-tracing approach to find the sensitivity and resolution of the parallelhole collimator by including the penetration of photons through the collimator due to the finite attenuation of the collimator material. "CSIM" can calculate the sensitivity of the collimator, the geometric and penetrating photon ratios, and the 1D and 2D point source response functions (PSF) with the statistical uncertainty for different hole shapes (e.g. square, hexagonal, and cylindrical). We have used "CSIM" to simulate the collimator of the YAP-(S)PETII small animal scanner. We present the analysis of the YAP-(S)PETII scanner round-hole parallel collimator designed for nuclear medicine imaging at 140keV. For this aim, different designs have been considered for a variety of source-collimator distances (b=5, 10, 15, 20cm). Resolution and sensitivity characteristics have been plotted as a function of the collimator thickness and the diameter of the hole. For each value of the source-collimator distance, and for each collimator thickness investigated, the trade-off between sensitivity and spatial resolution has been given as a series of characteristic curves. Then, we compare our simulated resolution and sensitivity results to the analytically calculated ones and found that the analytically calculated results for the YAP-(S)PETII scanner collimator are not far away the results predicted by CSIM and also with the experimentally measured resolution values.

Multiple-hit parameter estimation in monolithic detectors.

We examine a maximum-a-posteriori method for estimating the primary interaction position of gamma rays with multiple interaction sites (hits) in a monolithic detector. In assessing the performance of a multiple-hit estimator over that of a conventional one-hit estimator, we consider a few different detector and readout configurations of a 50-mm-wide square cerium-doped lutetium oxyorthosilicate block. For this study, we use simulated data from SCOUT, a Monte-Carlo tool for photon tracking and modeling scintillation- camera output. With this tool, we determine estimate bias and variance for a multiple-hit estimator and compare these with similar metrics for a one-hit maximum-likelihood estimator, which assumes full energy deposition in one hit. We also examine the effect of event filtering on these metrics; for this purpose, we use a likelihood threshold to reject signals that are not likely to have been produced under the assumed likelihood model. Depending on detector design, we observe a 1%-12% improvement of intrinsic resolution for a 1-or-2-hit estimator as compared with a 1-hit estimator. We also observe improved differentiation of photopeak events using a 1-or-2-hit estimator as compared with the 1-hit estimator; more than 6% of photopeak events that were rejected by likelihood filtering for the 1-hit estimator were accurately identified as photopeak events and positioned without loss of resolution by a 1-or-2-hit estimator; for PET, this equates to at least a 12% improvement in coincidence-detection efficiency with likelihood filtering applied.

A comparison of multislice computerized tomography, cone-beam computerized tomography, and single photon emission computerized tomography for the assessment of bone invasion by oral malignancies.

OBJECTIVE: The aim of this study was to compare the performance of cone-beam computerized tomography (CBCT) with multislice CT (MSCT) and single photon emission CT (SPECT) in the detection of bone invasion from oral malignancies. STUDY DESIGN: In this prospective investigation, 77 patients with histologically proven malignancy of the oral cavity received MSCT, CBCT, and SPECT imaging of the head presurgically. Radiologic evaluations were compared with histopathologic examinations of the resected tumor specimens. Receiver operating characteristic (ROC) analysis as well as the sensitivity, specificity, and positive and negative predictive values were calculated. RESULTS: The sensitivity, specificity, positive predictive value, and negative predictive value for MSCT were 0.8, 1.0, 1.0, and 0.75, respectively; for CBCT 0.92, 0.965, 0.98, and 0.875; and for SPECT 0.91, 0.4, 0.7, and 0.75. ROC analysis showed area under the curve values of 0.894 (95% confidence interval [CI] 0.806-0.982) for MSCT; 0.931 (95% CI 0.835-1.000) for CBCT, and 0.716 (95% CI 0.566-0.866) for SPECT. CONCLUSION: CBCT is accurate in predicting malignancies' bone involvement and can compete with MSCT and SPECT in detecting bone invasion in patients with oral malignancies.

Variations in PET/CT methodology for oncologic imaging at U.S. academic medical centers: an imaging response assessment team survey.

UNLABELLED: In 2005, 8 Imaging Response Assessment Teams (IRATs) were funded by the National Cancer Institute (NCI) as supplemental grants to existing NCI Cancer Centers. After discussion among the IRATs regarding the need for increased standardization of clinical and research PET/CT methodology, it became apparent that data acquisition and processing approaches differ considerably among centers. To determine the variability in detail, a survey of IRAT sites and IRAT affiliates was performed. METHODS: A 34-question instrument evaluating patient preparation, scanner type, performance approach, display, and analysis was developed. Fifteen institutions, including the 8 original IRATs and 7 institutions that had developed affiliate IRATs, were surveyed. RESULTS: The major areas of variation were (18)F-FDG dose (259-740 MBq [7-20 mCi]) uptake time (45-90 min), sedation (never to frequently), handling of diabetic patients, imaging time (2-7 min/bed position), performance of diagnostic CT scans as a part of PET/CT, type of acquisition (2-dimensional vs. 3-dimensional), CT technique, duration of fasting (4 or 6 h), and (varying widely) acquisition, processing, display, and PACS software--with 4 sites stating that poor-quality images appear on PACS. CONCLUSION: There is considerable variability in the way PET/CT scans are performed at academic institutions that are part of the IRAT network. This variability likely makes it difficult to quantitatively compare studies performed at different centers. These data suggest that additional standardization in methodology will be required so that PET/CT studies, especially those performed quantitatively, are more comparable across sites.

Variations in clinical PET/CT operations: results of an international survey of active PET/CT users.

UNLABELLED: This study gathered information about clinical PET/CT operations worldwide to help guide discussions on the use and standardization of clinical PET/CT. METHODS: A Web-based survey of PET/CT users was initiated in November 2009 through e-mail advertising using Academy of Molecular Imaging databases. Recipients were asked 58 questions related to demographics (e.g., location, number of PET/CT systems, and staffing), PET/CT operations and use, and variations in (18)F-FDG oncology imaging protocols. RESULTS: The responders were from centers in the Americas (71%), Europe (22%), Asia-Pacific (6%), and Middle East (1%), with most responding sites representing public health care institutions (60%). PET/CT systems were most frequently installed in nuclear medicine departments (59%). Of the sites operating a PET/CT system, 16% had 10 y or more of stand-alone PET experience. About 40% of all sites operated at least 2 PET/CT systems. PET/CT was most frequently used for applications in torso or whole-body oncology (87%), radiation therapy planning (4%), cardiology (4%), and neurology (5%). The average interval of fasting before an (18)F-FDG PET/CT examination was 7 ± 3 h (range, 4-12 h). Blood glucose levels were measured at 99% of sites, but acceptable maximal glucose levels varied substantially (an upper limit of 200 mg/dL was applied at >50% of the institutions). A weight-based radioactivity dose injection was performed at 44% of sites. The mean (18)F-FDG activity injected was 390 MBq (range, 110-585 MBq) for 3-dimensional PET of a 75-kg patient. The mean uptake time was 64 ± 14 min (range, 20-90 min). Split protocols involving patient repositioning and adapted imaging parameters were used at 51% of sites. Only 41% used patient positioning aids. Intravenous or oral CT contrast material was used at 52% of sites in up to 25% of patients. Most sites (90%) measured maximum standardized uptake value as an index of tissue glucose use. Only 62% of sites provided a fully integrated PET/CT report. CONCLUSION: An international survey among clinical PET/CT users revealed significant variations in standard (18)F-FDG PET/CT protocols. This finding illustrates the need for continuous training and ongoing standardization in an effort to optimize PET/CT in oncology.

Adaptive imaging for lesion detection using a zoom-in PET system.

Positron emission tomography (PET) has become a leading modality in molecular imaging. Demands for further improvements in spatial resolution and sensitivity remain high with growing number of applications. In this paper we present a novel PET system design that integrates a high-resolution depth-of-interaction (DOI) detector into an existing PET system to obtain higher-resolution and higher-sensitivity images in a target region around the face of the high-resolution detector. A unique feature of the proposed PET system is that the high-resolution detector can be adaptively positioned based on the detectability or quantitative accuracy of a feature of interest. This paper focuses on the signal-known-exactly, background-known-exactly (SKE-BKE) detection task. We perform theoretical analysis of lesion detectability using computer observers, and then develop methods that can efficiently calculate the optimal position of the high-resolution detector that maximizes the lesion detectability. We simulated incorporation of a high-resolution DOI detector into the microPET II scanner. Quantitative results verified that the new system has better performance than the microPET II scanner in terms of spatial resolution and lesion detectability, and that the optimal position for lesion detection can be reliably predicted by the proposed method.

Evaluation of an uncollimated printed paper transmission source used under scatter limiting conditions.

Transmission sources used for image attenuation correction, allowing image quantification, are collimated to reduce scatter. We propose the same effect can be achieved for an uncollimated source by increasing source to patient distance. The aim was to compare planar image performance characteristics and absorbed doses of uncollimated and collimated radioactive printed paper transmission sources. The scatter contribution to the uncollimated (99m)Tc source data was evaluated for different combinations of detector phantom distance, detector source distance and phantom source distance. Measurements were performed by increasing the Lucite phantom thickness in 1cm steps to 20 cm. Spatial resolution, detection efficiency and entrance absorbed dose rate were measured for the uncollimated and collimated transmission source images. Results derived from the energy spectra, obtained with the uncollimated transmission source indicate that scatter contribution increases with decreasing detector source distance. The scatter component in the uncollimated transmission images (detector source distances ≥ 60 cm; phantom source distances ≥ 40 cm) was comparable to that obtained with collimated transmission images. Attenuation coefficients obtained compared well (0.168 cm⁻¹ vs. 0.171 cm⁻¹). The full widths at half maxima differed by less than 0.9 mm. The detection efficiency of the uncollimated source was 2.5 times higher than obtained with the collimated source. The entrance absorbed dose obtained from an uncollimated source was 3.75 times larger than that obtained from the collimated source. An uncollimated transmission source (detector source distance ≥ 60 cm) results in acceptable image characteristics and presents a low cost, low dose, high efficiency option for transmission imaging.

Incorporation of a left ventricle finite element model defining infarction into the XCAT imaging phantom.

The 4D extended cardiac-torso (XCAT) phantom was developed to provide a realistic and flexible model of the human anatomy and cardiac and respiratory motions for use in medical imaging research. A prior limitation to the phantom was that it did not accurately simulate altered functions of the heart that result from cardiac pathologies such as coronary artery disease (CAD). We overcame this limitation in a previous study by combining the phantom with a finite-element (FE) mechanical model of the left ventricle (LV) capable of more realistically simulating regional defects caused by ischemia. In the present work, we extend this model giving it the ability to accurately simulate motion abnormalities caused by myocardial infarction (MI), a far more complex situation in terms of altered mechanics compared with the modeling of acute ischemia. The FE model geometry is based on high resolution CT images of a normal male subject. An anterior region was defined as infarcted and the material properties and fiber distribution were altered, according to the bio-physiological properties of two types of infarction, i.e., fibrous and remodeled infarction (30% thinner wall than fibrous case). Compared with the original, surface-based 4D beating heart model of the XCAT, where regional abnormalities are modeled by simply scaling down the motion in those regions, the FE model was found to provide a more accurate representation of the abnormal motion of the LV due to the effects of fibrous infarction as well as depicting the motion of remodeled infarction. In particular, the FE models allow for the accurate depiction of dyskinetic motion. The average circumferential strain results were found to be consistent with measured dyskinetic experimental results. Combined with the 4D XCAT phantom, the FE model can be used to produce realistic multimodality sets of imaging data from a variety of patients in which the normal or abnormal cardiac function is accurately represented.

A parallel excitation based fluorescence molecular tomography system for whole-body simultaneous imaging of small animals.

Challenges remain in imaging complete dynamic physiological processes in vivo through the whole small animal body using fluorescence molecular tomography (FMT). In this article, a novel non-contact full-angle FMT system that enables whole-body simultaneous imaging of small animals is presented. The whole-body simultaneous imaging ability is achieved by employing a line-shaped parallel excitation source, which can provide extended spatial sampling dataset to reconstruct multiple fluorescent targets distributed in whole animal body during one full-angle FMT imaging process. The key performances of this system were evaluated by a series of experiments. Quantitation linearity for over two orders of magnitude of fluorescence markers concentration was demonstrated, and an accessible simultaneous imaging domain of 4.0 x 1.5 cm² could be achieved utilizing the parallel excitation pattern. Moreover, the in vivo imaging feasibility and performance were validated by localizing two fluorescent targets implanted at different positions of a nude mouse. The results suggest that compared with conventional single point excitation FMT system, the proposed system can achieve a whole-body simultaneous imaging domain and impart the ability to image complete dynamic physiological processes in vivo.

Trends in PET scanner ownership and leasing by nonradiologist physicians.

PURPOSE: The aim of this study was to examine growth trends in ownership or leasing of private-office PET scanners by nonradiologist physicians. MATERIALS AND METHODS: The Medicare Part B Physician/Supplier Procedure Summary Master Files for 2002 through 2007 were used to collect the following data for each PET-related Current Procedural Terminology((R)) code: 1) annual procedure volume, 2) places of service for the procedures, and 3) specialties of the physicians filing the claims. To determine ownership or leasing, only technical and global claims that occurred in the nonhospital, private-office setting were included in the study. Professional component-only claims were not included. Procedure volume and growth trends were compared between radiologists and other specialties. RESULTS: Between 2002 and 2007, radiologist-owned Medicare PET scans increased by 259%, whereas nonradiologist-owned or nonradiologist-leased scans grew by 737%. Five specialty groups accounted for 95% of all nonradiologist PET volume in 2007: internal medicine subspecialties (28,324 studies in 2007), medical oncology (14,320 studies), cardiology (13,724 studies), radiation oncology (9,563 studies), and primary care (2,398 studies). In 2002, of all Medicare PET examinations performed on units owned or leased by physicians, the share for nonradiologists was 13%; their share rose to 24% in 2007. CONCLUSION: Although a large percentage of PET scans in private offices are done by radiologists, the growth rate among nonradiologists was far higher between 2002 and 2007 (259% for the former, 737% for the latter). The disproportionately rapid growth of PET scans performed on units owned by nonradiologists raises concern about self-referral at a time when policymakers are struggling to contain costs and reduce radiation exposure.