In vivo EPR tooth dosimetry for triage after a radiation event involving large populations

Abstract

The management of radiation injuries following a catastrophic event where large numbers of people may have been exposed to life-threatening doses of ionizing radiation will rely critically on the availability and use of suitable biodosimetry methods. In vivo electron paramagnetic resonance (EPR) tooth dosimetry has a number of valuable and unique characteristics and capabilities that may help enable effective triage. We have produced a prototype of a deployable EPR tooth dosimeter and tested it in several in vitro and in vivo studies to characterize the performance and utility at the state of the art. This report focuses on recent advances in the technology, which strengthen the evidence that in vivo EPR tooth dosimetry can provide practical, accurate, and rapid measurements in the context of its intended use to help triage victims in the event of an improvised nuclear device. These advances provide evidence that the signal is stable, accurate to within 0.5 Gy, and can be successfully carried out in vivo. The stability over time of the radiation-induced EPR signal from whole teeth was measured to confirm its long-term stability and better characterize signal behavior in the hours following irradiation. Dosimetry measurements were taken for five pairs of natural human upper central incisors mounted within a simple anatomic mouth model that demonstrates the ability to achieve 0.5 Gy standard error of inverse dose prediction. An assessment of the use of intact upper incisors for dose estimation and screening was performed with volunteer subjects who have not been exposed to significant levels of ionizing radiation and patients who have undergone total body irradiation as part of bone marrow transplant procedures. Based on these and previous evaluations of the performance and use of the in vivo tooth dosimetry system, it is concluded that this system could be a very valuable resource to aid in the management of a massive radiological event.

Introduction

The management of radiation injuries following the detonation of an improvised nuclear device (IND) or any other catastrophic event where large numbers of people may have been exposed to life-threatening doses of ionizing radiation will rely critically on the availability and use of suitable biodosimetry methods (Coleman et al. 2009). Unlike smaller incidents, where a limited number of affected people can be cared for effectively by intact medical resources, a massive event is expected to lead to a situation where the number of individuals involved would severely overwhelm any available conventional emergency medical resources; there also is a high likelihood that some local medical facilities will be incapacitated (Planning Guidance for Response to a Nuclear Detonation 2010). In such cases, information provided by appropriate biodosimetry tools, combined with suitable triage systems, would enable medical care to be provided most effectively to individuals who would be expected to benefit and allow decisions to be made by response managers to make most efficient use of the limited medical resources (Flood et al. 2011).

The US federal government, to guide preparations for effective responses to a major nuclear event, describes scenarios that would potentially expose very large numbers of people (Planning Guidance for Response to a Nuclear Detonation 2010). One such event, the denotation of a 10-kiloton IND, would likely impact a region as large as 7,500 km² and severely damage the infrastructure within a radius of up to 5 km from the epicenter. It would cause radiation, blast, and thermal injuries to the people within the affected area, with more serious combined injuries expected for those in closer proximity to the epicenter (Buddemeier and Dillon 2013). Fallout will be distributed over a wider region, and its impact will depend on environmental circumstances and vary with changing weather conditions. Fallout patterns will initially be difficult to determine and later be better defined by weather modeling and geographic radiation monitoring data. Consistent with the federal guidelines for such an event, it is most likely to occur in an urban area, resulting in hundreds of thousands of people potentially exposed to life-threatening ionizing radiation and other effects of the blast. A large number of individuals located in the fallout region may be instructed to shelter in place while others in the region will self-transport or be transported away from affected areas, with uncertain final disposition. Even where buildings remain, the infrastructure of transportation, communications, food distribution, and health care will be severely compromised within the region.

Across the affected populations, uncertainty of individual exposures will necessitate the application of biodosimetry to guide medical care, provide assurance to those who did not receive critical exposures, and provide information related to long-term effects of exposures. There are a number of biodosimetry techniques, including both biologically based and physically based methods, which can be considered for triage and guidance of medical care in such a scenario (Ainsbury et al. 2011; Swartz et al. 2011; Alexander et al. 2007; Swartz et al. 2014a). There are a number of critical characteristics that would be required to enable biodosimetry techniques to be appropriate for these needs, including applicability within an appropriate window of time after the detonation, rapid assessment of subjects to guide and enable medical intervention, suitable accuracy and precision, and independence from the influence of confounding factors (Swartz et al. 2011; Flood et al. 2011). At the initial stage of triage, the most important function of biodosimetry will be to quickly identify individuals likely to have received doses >2–3 Gy where acute medical care would affect survival. At this stage, a low false-negative rate is required, such that people in need of care would not be overlooked, and a modest false-positive rate may be acceptable. Subsequent stages of triage, which can benefit from the initial triage having successfully identified an expected large proportion of people who do not need immediate care, may employ more time-consuming or laborious methods on those identified as warranting further attention, which may therefore be able to provide greater precision.

The present paper focuses on in vivo electron paramagnetic resonance (EPR) tooth dosimetry, based on its having a number of valuable and some unique characteristics and capabilities that are particularly well suited for triage of a very large number of people (Swartz et al. 2010, 2011; Williams et al. 2011a, b; Flood et al. 2011). The fundamental basis for EPR tooth dosimetry is that radiation generates stable carbonate radicals within the calcium hydroxyapatite matrix of tooth enamel, and the relative density of these radicals can be measured using EPR and related to absorbed dose via an empirical calibration. Details of EPR tooth dosimetry, with emphasis on basic properties and in vitro studies, were recently reviewed by Fattibene and Callens (2010). EPR tooth dosimetry using isolated tooth enamel is an accepted and validated biodosimetry method that has been applied successfully for retrospective dose assessments following mass exposures including Chernobyl, Hiroshima, and Nagasaki, and nuclear weapons production and testing (IAEA2002; Ishii et al. 1990; Nakamura et al. 1998). Recent development of in vivo EPR tooth dosimetry has been motivated by the proliferation of nuclear capabilities and the threat of large-scale terrorist acts where there may be an urgent need to perform triage of large numbers of individuals potentially at risk for acute radiation syndrome and in need of care. In vivo EPR tooth dosimetry provides a physical dose estimate, with associated uncertainty, that can be combined with information available based on other assays and then compared to established dosimetric and general health thresholds to guide care. Alternatively, EPR tooth dosimetry results could be formatted to provide an estimate of the likelihood of an absorbed dose greater than a defined threshold. For triage applications, in contrast to more general retrospective biodosimetry, the ability to take measurements noninvasively, without the need to remove teeth or enamel samples, could enable rapid throughput and would eliminate the added steps and potential trauma associated with tissue sampling. Other advantageous features of in vivo EPR tooth dosimetry and additional considerations include

• Measurements are based on an endogenous local reporter (teeth) present in a majority of the population.

• Measurements with permanent upper incisor teeth (as being developed) are applicable for adults and older children, and can be adapted for measurements in younger children.

• Measurements provide a specific dose estimate (at the tooth site) for each subject.

• Measurements are noninvasive, requiring no tissue sampling.

• Measurements are nondestructive and do not affect the results of subsequent measurements, thus enabling repeated measurements to be made, e.g., when there is a need for increased precision or confirmation.

• The EPR signal in tooth enamel reflects the cumulative absorbed dose.

• The EPR tooth signal is unaffected by dose rate.

• The EPR tooth signal is directly sensitive to X-rays, gamma-rays, and charged particles; it is indirectly sensitive to neutrons via secondary radiation from the surrounding tissues.

• The radicals giving rise to the EPR signal are stable, allowing measurements to be taken at any time after irradiation. As described herein, a small fraction of radicals decay in the first hours after exposure, an effect that may be addressed if measurements are acquired a few hours after exposure.

• In vivo EPR tooth dosimetry can provide an estimate of the absorbed dose immediately, at the time and location of measurement.

• It is based on a physical process that is independent of confounding biologic factors and concomitant injuries, such as stress, physical trauma, and preexisting medical conditions and associated therapies.

• The technology is being developed to be field deployable.

• The technology is being developed for use by minimally trained individuals.

• EPR tooth dosimetry is complementary with other dosimetric assays. EPR tooth dosimetry measurements do not affect other assays, and the physical dose estimates provided may be combined with the results of biological assays to ascertain health risks for individual subjects.

Recent reports have described the technology developed for in vivo EPR tooth dosimetry and demonstrated the feasibility of providing accurate dose estimates for individual subjects with a deployable system (Williams et al. 2011a, b). The present report focuses on recent advances in the technology, which strengthen the evidence that in vivo EPR tooth dosimetry can make practical, accurate, and rapid measurements in the context of its intended use to help triage victims in the event of an IND. These advances provide evidence that the signal is stable and accurate to within 0.5 Gy, and that the method can be successfully carried out in vivo.

Materials and methods

EPR tooth dosimetry instrumentation

EPR tooth dosimetry is a magnetic resonance technique, where the resonant absorption of radiofrequency energy by radiation-induced radicals in teeth under the influence of an external magnetic field is measured and related to absorbed dose. Both in vivo and in vitro dosimetry measurements are taken using continuous-wave EPR with an operating frequency of 1.15 GHz and an associated magnetic field strength of 41 mT and employ a surface loop resonant coil for the measurements of the upper incisor teeth (Williams et al. 2011b). The instrument required to take these measurements consists of several major components, which include the magnet, a radiofrequency source and bridge, a resonant coil for signal detection, and appropriate mechanical structures to provide support and positioning capabilities necessary for in vivo measurements (Walczak et al. 2005; Salikhov et al. 2005; Williams et al. 2011b). We have produced and tested a prototype of a deployable EPR tooth dosimeter, comprised of a combination of commercial and custom-built components, which allows intact teeth in individual subjects to be measured through a completely noninvasive procedure (Fig. 1). This system utilizes about 30 kg permanent dipole magnet with a 17 cm gap produced by Resonance Research, Inc. (Billerica, MA, USA).

Fig. 1.

EPR tooth dosimeter prototype including permanent magnet, surface loop resonator detector, and ergonomic subject positioning assembles, and an instrument rack that contains the RF bridge, power supplies, and computer required to control the system and provides dosimetric information to the operator

The electronics for EPR detection and magnetic field sweeping are contained in a single deployable instrument rack that can be powered using the public electric supply or an electric generator. The radiofrequency bridges are custom designed and built within our laboratory to meet specifications required for maximal sensitivity and operation under the intended emergency conditions. Critical requirements include a low noise variable frequency source, computer control, and circuitry to enable automated tuning. The dosimeter and the data acquisition process are controlled using software developed in LabVIEW, which guides the operator through steps for tuning and collection of EPR data as described below. This software controls instrumental settings for the bridge, lock-in amplifier, and sweep and modulation field systems.

Surface loop resonators are employed to detect the radicals in the enamel. These resonators are built along conventional design, but with detection loops specifically tailored for dosimetric measurements of adult upper incisor teeth (Salikhov et al. 2003; Walczak et al. 2005; Williams et al. 2011b). Considerations in the loop design include the definition of the sensitive volume for the measurement and elimination of magnetic contamination, which can lead to noise and unwanted signals. The loops are constructed of high-purity silver wire, typically with an outer diameter of 10.4 mm and a wire thickness of 1.2 mm. Alternate loop geometries are under investigation to provide increases in dosimetric sensitivity and to minimize variation in measurements across subjects.

It should be noted here that in vivo EPR tooth dosimetry needs to be performed at a frequency that has no adverse effects when a person is measured and which is optimized to accommodate the need to assess a person, hence the technique reported here differs from in vitro EPR techniques based on crushing extracted teeth to obtain a sample of enamel to measure. However, much of early development and continuing assessments of instrumental improvements for in vivo application are done first in vitro using models and procedures that directly relate the results to the measurements taken in vivo. For in vitro applications, a single extracted tooth, or a mouth model with several extracted teeth placed in anatomically correct alignment, is placed in the same position within the magnet as is the human mouth during measurements. The use of in vitro models enables one to carry out the studies of dose–response, effects of anatomical variations, microwave properties of the mouth, etc. These in vitro models using human teeth avoid the complications that would occur if animal models were used because the teeth and mouths of animals are very different than those of human subjects.

The complete dosimetry system can be transported in two rugged Pelican-style boxes and put into operation in approximately 20 min. Individual measurements can currently be taken in 15 min or less and provide immediate estimates of absorbed dose. This prototype is appropriate for use by expert or nonexpert operators trained using established measurement protocols. In the state-of-the-art instrument, each component has been refined to maximize detection sensitivity and reproducibility, facilitate automated operation of the instrument appropriate for emergency use in the field, and provide comfortable and reliable subject positioning.

Measurement procedures

Protocols have been developed for taking measurements on unirradiated and irradiated volunteer subjects. The protocols establish methods for subject eligibility and positioning, proper placement of the surface loop resonator, maintaining proper hygienic conditions, and data acquisition and analysis. For each measurement session, subjects are instructed to sit facing the magnet and the height of the magnet is adjusted to provide optimal comfort; the subject’s head is then positioned in the magnetic field using a device to rest the upper teeth and support the upper lip and a strap to help immobilize the head during measurement. The sensing loop of the resonator is covered by a thin plastic hygienic barrier and placed directly on the tooth of interest, centered laterally with its upper edge placed 1 mm below the gum line. The pitch of the detection loop is adjusted to maintain contact of the upper edge of the resonator on the tooth while minimizing the separation between the tooth and the rest of the loop. The position of the loop is maintained using a lockable articulating arm and a spring-loaded assembly that ensures continuous contact between the tooth and the loop. The radiofrequency bridge is tuned via computer-controlled adjustment of the source frequency and phase, and the coupling between the bridge and resonator is optimized via manual adjustment of an inductive coupling assembly within the resonator. EPR data are then collected in a series of twenty 3-s scans of the magnetic field. To produce a measurement result, these scans are combined using a point-by-point median across scans to eliminate noise spikes, and this median spectrum is analyzed using a nonlinear least-squares fitting procedure to estimate the amplitude of the radiation-induced signal (RIS). This process is repeated to provide 5 RIS estimates, with independent positioning of the resonator on the tooth, tuning, and coupling to achieve independent spectral noise patterns, which can be ameliorated via averaging. These five estimates of the RIS amplitude are then averaged, and the result is converted to absorbed dose via an empirically based calibration curve that reflects the tooth type and instrumental configuration (Fig. 2).

Fig. 2.

In vivo measurement process for EPR tooth dosimetry. Data are for illustrative purposes only. a For each subject, 5 min of data are collected, consisting of 5 independent sets. b Twenty 3-s scans of the tooth and reference signals are collected per set. c The median spectrum is analyzed for each set using model-based spectral fitting, and the dosimetric signal amplitude is quantified. d Five median spectra and fits are collected per dosimetry session. e The average amplitude of the dosimetric signal (y-axis) is calculated with an associated uncertainty estimate. f The average amplitude of the dosimetric signal is converted to absorbed dose via an a priori calibration curve (above). See conversion on the right of e

Temporal stability following irradiation

The behavior of the RIS in tooth enamel immediately following irradiation was assessed via direct continuous measurements of the signal in isolated incisor teeth over 24 h. Continuous measurements for five central, upper incisor teeth were acquired and used to estimate the amplitudes and rates of any changes in the RIS. Measurements were taken continuously following irradiation for 24 h for four teeth and for 12 h for one tooth, with standard EPR data acquisition parameters but without repositioning of the resonator during the measurement period. All teeth were irradiated to an added dose of 30 Gy using a ¹³⁷Cs irradiator with a dose rate of 1 Gy/min. EPR measurements began within 15 min of the completion of irradiation. For each tooth, median spectra were calculated for each consecutive group of 20 scans and the radiation-induced EPR signal amplitudes were estimated using spectral fitting. These amplitudes were then fit as a function of time to an exponential decay function of the form $f(t)=a{e}^{-\frac{\ln (2)t}{b}}+c$ where $f(t)$ is the time-dependent radiation-induced EPR signal amplitude, a is the amplitude of the decaying component, b is the half-life of the decay, and c is the equilibrium amplitude, i.e., the amplitude once stable.

To examine the time immediately following the first 24 h for further evidence of stability, sets of spectra with repositioning of the resonator detection loop were acquired at four points in time, immediately following irradiation and 24, 48, and 72 h thereafter. (Methods and instrumentation were otherwise identical to the measurements taken continuously following irradiation to 30 Gy.) Repositioning of the resonator during the collection of each set of data enables averaging to reduce the impact of set-to-set baseline variability and provides more precise dosimetric data.

In order to document the long-term stability of the radiation-induced EPR signal, a single tooth that was irradiated to an added dose of 30 Gy was measured repeatedly over the course of 5 weeks. Measurements were taken according to the standard method for in vitro tooth measurements, where six sets of measurements are collected per session with independent positioning of the resonator and the average value of the amplitude of the radiation-induced EPR signal is calculated. A total of 41 sessions of measurements were acquired in the 5 weeks.

Dose estimation with mouth model

Dosimetry measurements were taken on extracted human upper central incisors mounted within a simple anatomic mouth model. Five models were created and measured, each consisting of four upper incisor teeth (two central and two laterals) placed in natural positions in an epoxy support structure mimicking the human jaw. For each model, the two central incisors (from the same donor) were separately measured. This arrangement allows the measurement of a tooth with both neighboring teeth irradiated to the same dose. Measurements were taken at six dose levels (0, 1, 2, 4, 6, and 10 Gy) to the teeth over 12 measurement days. The mouth models were irradiated individually using a ¹³⁷Cs irradiator calibrated for dose to water, with dose serially added on different days, to accumulate to the total doses. Two sessions of measurements were made on each mouth model daily for 2 days (total of four sessions at each dose). Each group was measured (at a given dose) in the afternoon of the first day and the morning of the second day, followed by irradiation. This pattern was then repeated for the other group at the same dose. This pattern allowed at least 24 h to elapse after irradiation until they were measured again at the higher dose. Spectra for each session were processed as described above, resulting in four estimates of the RIS for each tooth at each dose. Additional physical data were collected for each tooth to quantify the size of the tooth and characterize its curvature on central axial and sagittal planes. This information was used along with a characterization of the detection loop shape to apply a systematic, empirically based geometry correction for each tooth based on the linear dimensions and the volume of the enamel that extends into the detection loop. Estimated doses for each tooth at each dose are derived from the corrected signal amplitudes using a linear calibration based on the average native signal and dose–response observed across all teeth and doses.

Results

In vitro incisor measurements

Temporal stability following irradiation

Figure 3 shows the average response over the five teeth for 0–12 h based on the normalized data. In this plot, each set of measurements was normalized by dividing the data by its equilibrium value “c” prior to averaging, to take into account the different equilibrium EPR amplitudes following irradiation for each tooth used in this study, which depends on tooth size and the detected enamel volume. The averaged amplitude versus time was then fit to the same exponential model as above. The estimated parameters recapitulate the decay dynamics seen from the individual tooth analyses, with an average half-life of 3.6 h and a decay amplitude equal to 13 % of the equilibrium amplitude. According to these parameters, a tooth will on average undergo about six half-lives in 24 h, achieving ~99 % of its decay; assuming the amplitude of the decay is 15 %, it will have decayed to within 0.1 % of its equilibrium value by the end of the first 24 h. Thus, measurements taken in the teeth after the first 24 h are likely to have reached equilibrium; measurements for less than 24 h can be used after an appropriate correction factor, assuming that the interval between exposure and measurement can be estimated. This result is consistent with the literature and implies that measurements can begin well within the time frame when biodosimetric measurements for triage are likely to be initiated.

Fig. 3.

Average response of the radiation-induced EPR signal in incisor tooth enamel during the 12 h immediately following irradiation to a dose of 30 Gy. In vitro measurements were taken for five teeth, and data for each tooth were divided by its equilibrium value (t = ∞) to normalize data prior to averaging. The average response, ±standard error of the mean (SEM), and an exponential fit to the data are shown. EPR amplitudes are normalized to the equilibrium value. Parameter “b” is in unit of hours

Figure 4 shows the radiation-induced EPR signal amplitudes at these time points averaged across nine teeth. Prior to averaging, each set of measurements was normalized to the average value for the respective tooth over the final three time points (24, 48, and 72 h after irradiation). Based on these data, the amplitude of the decaying component was 20 % of the equilibrium value and the normalized amplitudes of the EPR signals (i.e., the % values of the equilibrium value) at the 24, 48, and 72 h time points were 1.02 ± 0.02, 1.00 ± 0.02, and 0.97 ± 0.03, respectively. Within the errors in these normalized amplitude measurements, these data indicate that the tooth signals remain stable after the first 24-h period post-irradiation.

Fig. 4.

Average response of the radiation-induced EPR signal in incisor tooth enamel at 24-h intervals immediately following irradiation to a dose of 30 Gy. In vitro measurements were taken for nine teeth, and data for each tooth were divided by the average of its 24, 48, and 72 h amplitudes to normalize data prior to averaging. Error bars denote the standard error of the mean (SEM)

The measurements for each week were averaged together to assess long-term stability of the radiation-induced EPR signal. The mean signal amplitude and SEM were calculated for the measurements acquired each week, as shown in Fig. 5. The average relative amplitude of the radiation-induced EPR signal was 0.58 ± 0.01 (mean ± SD) and had a slope of 0.0 units/week with a 95 % confidence interval of −0.01 to +0.01.

Fig. 5.

The long-term stability of the radiation-induced EPR signal in tooth enamel is shown in measurements acquired over a 5-week period for an incisor tooth that had been irradiated to 30 Gy. Response of the radiation-induced EPR signal in tooth enamel over a 5-week period, at 24-h intervals immediately following irradiation of an incisor to a dose of 30 Gy. The measurements for each week were averaged together, and SEM values were calculated. EPR amplitudes are not normalized in this figure

Dose estimation with mouth model

When averaged across the four measurement sessions, the standard error of inverse dose prediction (SEIP) (Demidenko et al. 2012) was observed to be 0.58 Gy across all 10 teeth and six doses. For homoscedastic populations, where the variance is independent of dose, populations, SEIP equals the critical level with α = β = 0.16 (Currie 1968). When outlier effects were taken into account using bootstrap methods to serially examine the impact of using nine teeth to estimate the dose for the 10th, the average SEIP dropped to 0.50 Gy, which is arguably the threshold for variations that are clinically significant (Fig. 6). When the geometry correction factor was omitted, the SEIP was 1.24 Gy, suggesting that geometry is an important adjustment to include. This evidence supports earlier findings, using measurements of molar teeth with surface loop resonators, that geometry corrections are important for improving the estimates (Iwasaki et al. 2005).

Fig. 6.

EPR dosimetry measurements (y-axis) taken for ten central incisors that were serially irradiated in simple anatomic mouth models including neighboring lateral incisor teeth. The 95 % confidence limits for independent measurements (CLI) and mean response values (CLM) are provided. All doses are given in Gy

In vivo measurements

Throughout the course of development, evaluations of in vivo EPR tooth dosimetry technology and procedures in human subject measurements were carried out, including the measurements of unirradiated subjects and patients who have received radiation to their oral cavities during radiation therapy. We have also included regular deployment exercises as an integral part of our overall effort (Nicolalde et al. 2010; Williams et al. 2011b). All measurements described in the present study were taken following Institutional Review Board review and approval of protocols.

Field deployment exercises

The tooth dosimetry system has been tested in several “field” simulation settings. Throughout the initial development phases, the focus on the intended use was maintained, including field deployment of the technology to a location nearby to the radiation disaster, the need for rapid and high throughput in order to assess large numbers of people, and operation by minimally trained operators. The ability to use the tooth dosimeter in the field has been evaluated in a series of deployment exercises, including operation at a local firehouse, an international EPR conference, at several annual Dartmouth cancer center fundraisers held in tents and using generator power, and in the lobby of the Dartmouth–Hitchcock Medical Center at an event memorializing the anniversary of the Fukushima nuclear power plant accident (Nicolalde et al. 2010; Williams et al. 2011b). At a 2011 fundraiser, 83 unirradiated volunteers were measured using two dosimeters over a period of approximately 12 h. During the 2012 Fukushima memorial event, 63 unirradiated volunteers were measured over 2 days. A throughput of 15 min per subject was observed in these and several earlier exercises. However, these exercises measurement procedures were not optimized for throughput, and each dose estimate was based on approximately 5 min of EPR data collection. These exercises have served several important purposes, including the evaluations of the instruments and measurement procedures under nonlaboratory conditions where unforeseen measurement conditions (including wind and rain), heterogeneities across subjects, and “real-world” operation of the instrument can be experienced and used to guide the efficient development of the technology for the intended use.

Unirradiated subject and total body irradiation patient measurements

An assessment of the use of upper incisor measurements for dose estimation and screening was performed with volunteer subjects who have not been exposed to significant levels of ionizing radiation and patients who have undergone total body irradiation as part of bone marrow transplant procedures. In vivo measurements were taken as described above. Recruitment is ongoing, and to date, almost 500 volunteers have been measured, of whom more than 30 are patient volunteers who have had radiation therapy that resulted in in vivo radiation to the teeth. The majority of these measurements were acquired in the course of instrument development, where different instrumental settings, components, and procedures were employed.

Figure 7 presents an analysis of measurements with 59 subjects, including both unirradiated volunteers and irradiated patients. Measurements of patients in this analysis include one for a patient who received a single prescribed fraction of 1.5 Gy, 12 in patients who received a single fraction of 2 Gy, and three measurements taken in one patient who received fractionated total body irradiation to a total dose of 12 Gy. Data for 45 unirradiated subjects are included. Since EPR measurements are nondestructive, noninvasive, and the signal is indefinitely stable, independent repeated measurements can be taken, as has been true for many of both unirradiated subjects and irradiated patients. The estimated RIS amplitudes were used to estimate an in vivo dose–response curve, in which the EPR signal in volts was related to the known dose (in Gy). These data demonstrate three important findings. First, consistent with in vitro results, the dose–response curve for measurements taken in vivo, including patients whose radiation occurred in vivo, is linear. The intercept of the observed linear dose–response, related to native and background signals, corresponds to a dose of approximately 2.5 Gy. Second, the SEIP based on these data is 1.25 Gy. Last, the standard deviations of the RIS amplitudes, for the 0, 2, and 12 Gy doses where multiple measurements with multiple subjects were acquired, appeared to be uniform across doses. This latter observation is consistent with having an additive instrumental noise source present, rather than being due to interpersonal variations in dose–response where a proportional dependence of the deviation on dose would be expected. This in turn is indicative of the need to improve precision by improving the instrument, rather than by correcting for interpersonal confounders. Moreover, based on similar experiences with in vitro measurements as described above, the SEIP is likely to be reduced significantly if a geometry correction factor to account for variations in incisor tooth size and enamel volume is applied in the future.

Fig. 7.

Fifty-nine in vivo measurements of unirradiated and irradiated volunteers using upper incisor teeth. For boxes at 0 and 2 Gy, the central mark is the median, the edges of the box are the 25th and 75th percentiles, and the whiskers extend to the most extreme data points. At 1.5 and 12 Gy crosses denote individual measurements. These data establish an in vivo calibration curve (EPR amplitude in V versus dose in Gy) with a 1.25 Gy standard error of inverse dose prediction (SEIP)

Discussion

Based on these and previous evaluations of the performance and use of the in vivo tooth dosimetry system, we are confident that in vivo EPR tooth dosimetry will be a very valuable resource to aid in triage decisions following a very large radiological event, such as the detonation of an IND. It is likely to be particularly useful during the first stage of triage, where onsite and rapid information for triage of individuals at risk for acute radiation syndrome is necessary. Initial triage is likely to prioritize the identification of subjects exposed to clinically significant doses, for example above 2 Gy, with attention paid to minimizing the false-negative rate to ensure that those in need of care proceed into later levels of triage and the medical care system as appropriate.

The stability over time of the amplitude of the RIS at L-band, 1.2 GHz, in tooth enamel was measured to confirm expectations based on scientific principles and evidence from the literature. At higher frequencies, e.g., X-band at near 10 GHz, the temporal dynamics of the observed RIS are complex due to the possibility to distinguish a variety of radical species with different g-values and decay properties. The distinct properties of these radicals are reviewed by Fattibene and Callens (2010), including the discussion of recombination and transformation decay processes and proposed transient radical effects that can lead to transient increases in peak-to-peak signal amplitudes in the hours following irradiation (Sholom et al. 1998). The study described here was carried out to directly observe the temporal signal dynamics at L-band and assess their impact on dosimetric analyses for triage purposes. For triage purposes, there potentially may be a need to measure any time from immediately after exposure, and EPR tooth dosimetry has the capability to meet this need (Swartz et al. 2014b). There also is a need for the measurements to provide a robust estimate of dose for periods up to a few weeks until the benefit from triage for acute radiation syndrome (ARS) becomes moot. In order to confirm the feasibility of meeting these needs, measurements were taken to assess both short-term and long-term stabilities within the time frames pertinent to biodosimetry for triage, i.e., focusing on the first 24 h, the first 72 h, and over a 5-week period. These measurements indicate that the radiation-induced EPR signals from exposed whole teeth measured at L-band are stable beginning 24 h after exposure, during the time frame that emergency triage would likely be carried out, and that decay corrections to address temporal instability in the first 24 h are feasible.

Current efforts for the refinement of the EPR tooth dosimetry system are focused on refining the existing capabilities for automated, or semiautomated, operation such that the instrument can be operated by minimally trained personnel, further minimization of the SEIP, and preparation for the FDA regulatory approval processes required for use of the system. In addition to these foci for the mature L-band dosimetry system, the use of higher excitation frequencies and the use of pulsed-mode EPR data collection as potential long-term improvements to the system are also investigated.

The existing system is already capable of semiautomated operation, where the tuning of the spectrometer is accomplished via computer control, and the role of operator includes the guidance of the subject into position for measurement, the manual placement of the resonator detection loop on the surface of the tooth to be measured, and initiation of the measurement process via a simple graphical user interface. A number of ergonomic refinements relative to earlier versions of the instrument were implemented, to ensure reliable positioning of the subject within the magnet, including the use of a bite plate that guides the subject into the proper position and a system of support structures that enables the subject to comfortably remain in this position during measurement. A prototype of a computer-controlled set of linear and rotational stages has been implemented to enable controlled positioning, or repositioning, of the resonator on the tooth surface. This system can be used to reposition the resonator for repeated measurements, as is done in our existing measurement protocol, alleviating the need for the operator to perform these repetitive actions and possibly increasing the precision of resonator replacement and throughput. If the need for fully automated resonator placement is established, there is potential for this positioning system to be coupled with a computer vision system to identify the tooth position and angle and thereby guide the initial resonator placement.

Improvements in dosimetric precision are being achieved via several pathways, including the identification of the most effective strategies for averaging of random and systematic noise sources, the incorporation of additional quality assurance and verification procedures for proper instrumental setup and resonator positioning, and refinement of resonator fabrication procedures to eliminate sources of noise and paramagnetic contamination. Additionally, systematic procedures have been instituted for the identification, and removal, of spectra that have been corrupted by noise, e.g., that might result from subject motion or unintended movement of the detection loop. A combination of these efforts has allowed a SEIP of 0.50 Gy for the in vitro mouth model study to be achieved, and these procedures are now being applied similarly for in vivo studies. In the mouth model study, in addition to identifying the importance of geometric adjustment, preliminary analyses suggested that the averaging of data collected in independent sessions, separated by some interval in time, is beneficial and improved the precision significantly. Given the near-absolute stability of the radiation-induced radical density, it is likely that this effect is driven by significant temporal coherence of baseline noise sources in the existing instrument or potentially related to systematic variations in manual resonator placement. To avoid the need to expand the time needed for data acquisition, other approaches to perturb the noise and allow effective averaging, such as using small fixed rotations of the magnet in between collection of serial data sets, are being investigated.

There are several fundamental advantages associated with increases in RF frequency and magnetic field strength for in vivo EPR tooth dosimetry. It is predicted that the signal-to-noise ratio will increase at least in proportion to increases in these values (Hutchison 1971; Rinard et al. 2002). The use of higher RF frequency results in an increase in the magnitude of dielectric losses, which limits the depth of penetration of the RF field into the sample being measured and improves definition of the volume of sensitivity within the enamel volume. Decreased RF wavelengths are predicted to support the development of smaller detection loops, which also provide improved definition of the sensitive volume and suppression of dosimetric variations associated with tooth size and enamel thickness. Another significant advantage of the use of higher-frequency EPR is an increase in the spectral separation between the radiation-induced EPR signal and that from radical species that are native to the tooth enamel but not dependent on irradiation. With such discrimination, variability in the native signal amplitude across subjects can be better isolated from the RIS amplitude estimate and enable more accurate dosimetry.

In order to verify these expectations, we have constructed an S-band dosimetry system operating at 2.4 GHz and utilizing a conventional electromagnet. Following basic testing of the functionality of the S-band system, a preliminary dosimetric comparison to the L-band system, using state-of-the-art optimized resonators and acquisition parameters for both systems, has been conducted. For both systems, data acquisition and spectral analysis procedures were similar to those described above. For all comparisons, a single incisor tooth was used with an absorbed dose of 10 Gy. Each measurement set consisted of 20 consecutive scans, each requiring 3 s, and 18 sets of measurements were taken with each system. For additional verification of the performance of the S-band system and assessment of its stability, these measurements were repeated after approximately 1 week. The different types of magnets could potentially affect the performance assessment of EPR dosimeters. Accordingly, comparative EPR measurements at L-band were taken with both the electromagnet and permanent magnet. The microwave bridge, resonator, tooth, lock-in amplifier, and data acquisition system were held constant, with only the magnet and modulation coils varying.

Data indicate that the signal-to-noise ratios (SNR) of the spectra are unaffected by the magnet and a slightly increased baseline distortion was observed using the electromagnet. Therefore, a measurement with an electromagnet does not appear to confer an advantage, and a comparison of the S-band system with electromagnet versus the L-band with permanent magnet will give a conservative estimation of the relative benefits of operation at S-band, with the potential to have even better performance when a permanent magnet is implemented for S-band. Representative 3-s scans acquired with each system are shown in Fig. 8. Analysis of the complete set of measurements indicates that SNR is approximately 2.5-fold better at S-band than at L-band, the reproducibility of dose estimation, with standard six reinstallations of the resonator for in vitro measurements, is at least 1.5-fold better than at L-band, and the capability to quantify doses <1 Gy is significantly enhanced. A similar improvement in SNR through averaging alone would require a >6-fold increase in data acquisition time. Based on these promising results, this S-band instrument is being further refined to enable in vivo measurements, including the addition of a magnet appropriate for the measurements of intact teeth and the incorporation of enhanced automation.

Fig. 8.

Comparison of L-band and S-band spectra acquired for a 10 Gy tooth with individually optimized conditions, as described. Each scan required 3 s for collection. The improvement in SNR (about 2.5-fold) is apparent

Conclusion

The development of in vivo EPR tooth dosimetry as described here is aimed to provide physically based dose estimates, with associated uncertainty, in individuals for the purposes of aiding triage decisions in the field following a radiation event involving large numbers of individuals. There are several characteristics of this approach that are particularly appealing for this intended use, including

• the direct detection of radiation-induced radicals in human tissue

• the unique stability of the radiation signature

• its noninvasive nature

• the opportunity to make repeated measurements

• the ability to provide immediate absolute estimates of absorbed dose

• the lack of confounding metabolic and biologic interactions

• its potential to act as a natural complement to biologically based techniques

The deployment exercises and performance evaluations described indicate that EPR tooth dosimetry can discriminate dose levels for triage (i.e., 2 Gy) (Grace et al. 2010) with less than 5 min of measurement time. With SEIP values of 0.5 and 1.25 Gy for in vitro and in vivo measurements, these data indicate that the current system and procedures can discriminate unirradiated subjects from those irradiated to 2 Gy with accuracies of approximately 98 and 80 %, respectively. Direct measurements of the stability of the RIS at L-band confirm that a small transient component exists. Valid dose estimates can be made anytime beginning within hours of exposure and even earlier with correction factors to account for the small decaying component. Ongoing developments are aimed at increasing overall throughput, improving the precision of dose estimation, and assuring reliable use of the system in the field by nonexpert operators. Additional developments are also being carried out to explore the use of higher RF frequencies and pulsed-mode detection as methods to further increase the sensitivity of the technique.