Kinetic analysis of PET data requires continuous measurement of radioactivity in the arterial blood throughout the acquisition time, termed the arterial input function. The arterial input function is used as an input to compartmental modeling, which can be a better predictor of disease progression than SUV measurements from static PET images. Current common methods of measuring blood concentrations include image-derived, population-based, and manual sampling. These all have challenges due to logistical and technologic issues, as well as patient burden. The aim of this study was to design, develop, and assess a device that is practical and effective for the routine measurement of β-emitting radiotracer concentration in blood without the drawbacks of current methods and for which metabolite analysis is not required. Designs that integrated a scintillating fiber and a silicon photomultiplier with a general-purpose venous access catheter for in vivo measurement were considered. Other design requirements included miniaturization, high sampling rates, and stopping power for β-particles. Preliminary prototypes were designed to test the feasibility of the concept. Phantom tests were developed to mimic human vasculature. Tests of linearity, sensitivity, signal-to-noise ratios, the impact of vein diameter, and the influence of γ-radiation were conducted. Prototype sensors were constructed using 2 different diameters of polystyrene-based scintillating fibers. Fibers were custom-polished and fixed to a silicon photomultiplier. Sensor output was linear, with = 0.999 over the range from 0.037 to 9.25 MBq/mL. Absolute sensitivity was approximately 450 counts per second per MBq/mL. Measured signal-to-noise ratios ranged from 1.2:1 to 3.2:1 using a blood-to-tissue concentration ratio of 1:1. Sensor output increased with vein diameter and showed no sensitivity to γ-radiation. In experiments with phantom models, the prototype provided accurate measurements of β-emitting radiotracer concentration. The design will be refined for in vivo testing. The ability to routinely gather blood input function data would facilitate the adoption of kinetic modeling of PET data.

PurposeEach year in the United States, approximately 18.5 million nuclear medicine procedures are performed. Various quality control measures are implemented to reduce image errors and improve quantification of radiotracer distribution. However, there is currently no routine or timely feedback about the quality of the radiotracer injection. One potential solution to evaluate the injection quality is to place a topical scintillation sensor near the injection site to record the presence of residual activity. This work investigates a sensor design for identification of injections where the prescribed radioactive activity is not fully delivered into the patient's circulation (an infiltration).MethodsThe sensor consists of a single unshielded bismuth germanate (BGO) crystal (3 mm × 3 mm × 3 mm). Using radioactive sources with gamma energies that span the range commonly used in nuclear medicine, we quantified energy resolution and linearity. Additionally, we computed sensitivity by comparing the calculated incident activity to the activity measured by the sensor. Sensor output linearity was calculated by comparing measured data against the radioactive decay of a source over multiple half‐lives. The sensor incorporates internal temperature feedback used to compensate for ambient temperature fluctuations. We investigated the performance of this compensation over the range of 15°C–35°C.ResultsEnergy spectra from four sensors were used to calculate the energy resolution: 67% for 99mTc (141 keV), 67% for 133Ba (344 keV), 42% for 18F (511 keV), and 32% for 137Cs (662 keV). Note that the energy used for 133Ba is a weighted average of the three photon emissions nearest to the most abundant (356 keV). Sensor energy response was linear with a difference of 1%–2% between measured and predicted values. Energy‐dependent detector sensitivity, defined as the ratio of measured photons to incident photons for a given isotope, decreased with increasing photon energy from 55.4% for 99mTc (141 keV) to 3.3% for 137Cs (662 keV). Without compensation, error due to temperature change was as high as 53%. Temperature compensation reduced the error to less than 1.4%. Sensor output linearity was tested to as high as 210 kcps and the maximum magnitude error was 4%.ConclusionsThe performance of the sensor was adequate for identification of excessive residual activity at an injection site. Its ability to provide feedback may be useful as a quality control measure for nuclear medicine injections.