Comparison of 3 γ-probes for simultaneous iodine-125-seed and technetium-99m breast cancer surgery: NEMA standard characterisation with extended processing

Purpose Iodine-125 (125I) seeds can be used as landmarks to locate non-palpable breast lesions instead of implanting metal wires. This relatively new technique requires a nuclear probe usually used for technetium-99m (99mTc) sentinel node detection. This study aimed to compare the performances of different probes and valid the feasibility of this technique, especially in the case of simultaneous 125I-seed and 99mTc breast cancer surgery. Methods Three probes with different features (SOE-3211, SOE-3214 and GammaSUP-II) were characterised according to the NEMA NU3-2004 standards for a 99mTc source and a 125I-seed. Several tests such as sensitivity, linearity or spatial resolution allowed an objective comparison of their performances. NEMA testing was extended to work on signals discrimination in case of simultaneous detection of two different sources (innovative figure of merit “Shift Index”) and to assess the 99mTc scatter fraction, a useful parameter for the improvement of the probes in terms of detector materials and electronic system. Results Although the GammaSUP-II probe saturated at a lower activity (1.6 MBq at 10 mm depth), it allowed better sensitivity and spatial resolution at the different NEMA tests performed with the 99mTc source (7865 cps/MBq and 15 mm FWHM at 10 mm depth). With the 125I-seed, the GammaSUP-II was the most sensitive probe (3106 cps/MBq at 10 mm depth) and the SOE-3211 probe had the best spatial resolution (FWHM 20 mm at 10 mm depth). The SOE-3214 probe was more efficient on discriminating 125I from 99mTc in case of simultaneous detection. The SOE probes were more efficient concerning 99mTc scatter fraction assessments. The SOE-3211 probe, with overall polyvalent performances, seemed to be an interesting trade-off for detection of both 125I and 99mTc. Conclusion The three probes showed heterogeneous performances but were all suitable for simultaneous 99mTc sentinel node and 125I-seed detection. This study provides an objective and innovative methodology to compare probes performances and then choose the best trade-off regarding their expected use.


Introduction
For the excision of non-palpable breast lesions, the surgeon needs a landmark previously placed under radiological or ultrasound imaging control. Guided surgery is an evolving field [1], whether it concerns fluorescence [2], ferromagnetic detection [3], radiofrequency [4] or radioactivity detected by γ-camera [5]. But in France, the gold standard technique remains implanting a metal wire from the lesion to the skin. The Antoine Lacassagne Centre (Nice, France) is the first French hospital authorised to offer an alternative solution with an iodine-125-seed ( 125 I-seed) implanted into the tumour [6]. The two techniques are currently compared within the randomised prospective clinical trial "IODINE BREAST" (NCT02759133). The literature already tips the scales in favour of the 125 I-seed technique [7]. Indeed, using the metal wire leads to more important margins and involves some discomfort for the patient. The iodine technique seems to be an answer to these issues with a lower rate of repeated surgery and a lower psychological impact [8,9]. However, specific nuclear probes are needed to detect the seed into the breast.
In the context of breast tumour surgery, it is common to inject technetium-99m ( 99m Tc)-nanocolloids around the areola to find and remove the sentinel node, using nuclear probes as well. The 125 I and 99m Tc radionuclides emit photons of 27 and 140 keV respectively; thus, 99m Tc scatter signal can impact on the detection of the 125 I-seed (Fig. 1). Fig. 1 Illustration of the EURORAD low (20-40 keV) and high (80-170 keV) energy windows onto the energy spectra of a 99m Tc source and a 125 I-seed in a scatter medium performed with the SOE-3214 probe. The sources depths (Z) were considered from the in vitro clinical simulations. These spectra were acquired over 30 s and are presented for 1 MBq activity. The 99m Tc scatter signal corresponds to the 99m Tc signal recorded in the low energy window The main purpose of our study was the characterisation of three different gamma probes for both sources of 99m Tc and 125 I according to the NEMA NU3-2004 international standards [10]. We aimed to go further exploiting data from two NEMA tests leading to two additional indexes: -The "Shift Index" was defined as an innovative figure of merit of signals discrimination in case of simultaneous detection of two different sources.
-The 99m Tc scatter fraction was calculated according to the source activity and depending on the spatial positioning of the probe. This parameter is useful to improve the probes, especially in terms of detector materials and electronic system.
The probes performances were compared in order to determine if one of them was more suitable to this technique, especially in terms of simultaneous detection of 125 Iseed and 99m Tc sources.

Material and methods
Probes (Table 1) Three gamma probes (without external collimator) from two companies with a quality control up to date were considered for this study: SOE-3211 and SOE-3214 (EURO-RAD, Strasbourg, France) and GammaSUP-II (CLERAD, Clermont-Ferrand, France).
SOE-3211 from EURORAD is a bent nuclear probe equipped with a cadmium telluride (CdTe) semiconductor detector designed to detect photons in the range 20-170 keV. Thus, it is dedicated to the detection of 125 I (27 keV) and 99m Tc (140 keV). The head of the probe is 11 mm diameter with a stainless-steel entrance layer. SOE-3214 from EURORAD is a straight opto-nuclear probe with the same detection features as the SOE-3211 probe. In addition to detect 125 I and 99m Tc signals, two optical fibres allow infracyanine green fluorescence for sentinel nodes detection. The head of the probe is 12 mm diameter with a resin entrance layer. GammaSUP-II from CLERAD is a straight nuclear probe equipped with a thalliumdoped caesium iodide [CsI(Tl)] scintillator coupled with a photomultiplier for the detection of photons in the energy range 20-400 keV. Thus, this probe can detect photons of low and high energies. The head of the probe is 12 mm diameter with a stainlesssteel entrance layer. For every NEMA standards testing, count rates were recorded in the clinical energy windows corresponding to the source: low energy window (20-40 keV for EURORAD and CLERAD) for the 125 I-seed and high energy window (80-170 keV for EURORAD and 120-170 keV for CLERAD) for the 99m Tc source. For the extended NEMA testing, (clinical simulations and 99m Tc scatter fraction), 99m Tc scatter signal was measured in the low energy window while performing the spatial resolution and count rate capability tests ( Table 2, Fig. 1).
NEMA testing (Fig. 2) Among every NEMA tests, we chose the seven most relevant ones for our study which were practically feasible: sensitivity in air, sensitivity in a scatter medium, sensitivity through side shielding in air, sensitivity to scatter, spatial resolution in a scatter medium, count rate capability in a scatter medium and side and back shielding. Each test was performed according to the NEMA NU3-2004 standards [11] with some adaptations for the in vitro clinical simulations and 99m Tc scatter fraction assessments.
99m Tc sources were prepared by pouring liquid drops in a plastic capsule filled with paper and clogged with wax to be sealed. The capsule was a cylinder of 2.5 ± 0.5 mm diameter by 1.5 ± 0.5 mm height and was assumed to be punctual (volume < 14 μL).
The activity was measured using a MEDI 404 (MEDISYSTEM, Guyancourt, France) dose calibrator with an accuracy of less than 5%. 125 I sources IsoSeed® I25.S06 were provided by Eckert & Ziegler BEBIG as cylinder seeds of 0.8 mm diameter × 4.5 mm height. We considered the activity from the calibration certificate taking into account the radioactive decay. The measurements were performed using a water tank 36.5 cm long × 21.4 cm large × 16.5 cm height, filled or not with water according to the tests and a Fisso arm to ensure a precise, stable and reproducible probe positioning.

Sensitivity in air
The source was fixed at the centre of the air tank at 12.5 cm height. The probe was aligned with the source along the vertical axis. Measurements were performed with the tip of the probe at 0 (contact), 10, 30 and 50 mm above the source.

Sensitivity in a scatter medium
The source was fixed at the centre of the water tank at 12.5 cm height. The head of the probe was aligned with the source along the vertical axis. Measurements were performed with the tip of the probe touching the water surface at 10, 30 and 40 mm above the source.

Sensitivity through side shielding in air
The probe was positioned at the centre of the wall of the air tank and aligned along the horizontal axis. The source was fixed 50 mm further on the same wall tank and at the same height.

Sensitivity to scatter
The setup was the same as the one used for the sensitivity through side shielding in air test at the exception the tank was filled with water. For a same probe and source, if the ratio between the count rate without and with scatter medium exceeded 10%, the sensitivity to scatter value was corrected by subtracting the sensitivity through side shielding in air value.

Spatial resolution in a scatter medium
From the setup used for the sensitivity in a scatter medium test, several measurements were performed through the X axis from − 50 to + 50 mm with the probe aligned according to the vertical axis at 10, 20 and 30 mm above the source. The measurement steps were adapted according to the distance from the X axis origin: 2.5 mm steps for 0 < |X| < 5 mm; 5 mm steps for 5 < |X| < 20 mm; 10 mm steps for 20 < |X| < 50 mm. The two measurements for the same distance from X axis origin (on X negative and X positive) were averaged in order to obtain symmetrical curves. Spatial resolution was expressed as the full width at half maximum (FWHM) and the full width at tenth maximum (FWTM) using linear interpolation around the 50% and 10% of the maximum respectively. and 50 mm) and sensitivity through side shielding in air assessments, a and b setups were respectively used without water. The probe was slided through the X axis for the spatial resolution test; it was fixed at X = 0 mm for every other tests

Count rate capability in a scatter medium
The source was fixed at the centre of the water tank at 12.5 cm height. The probe was aligned with the source along the vertical axis. Measurements were performed with the tip of the probe touching the water surface, 10 mm above the source. For 99m Tc, measurements were performed over 48 h to obtain count rate values for an activity ranging from 0.1 to 7 MBq. For 125 I, we used different seeds to obtain count rate values for an activity ranging from 1 to 10 MBq. The count rate capability corresponded to the activity value above which the loss of count rate linearity exceeded 20%.

Side and back shielding
The source was slowly moved around the probe's head surface and in contact with it. It was ensured the source did not shine directly the detector from the tip of the probe. The highest count rate and the source position associated were recorded. From this test and the sensitivity in air, the shielding effectiveness (SE) was calculated (Eq. 1).

SE ¼
Sensitivity in air contact −Side and back shielding Sensitivity in air contact Â 100 ð1Þ For the sensitivity, spatial resolution and count rate capability tests (in air and scatter medium), the probe position was adjusted until an equivalent count rate at ± 15 mm on the X and Y axis was measured. This methodology ensured to be at the maximum count rate at the origin of the XY plane.
For every test, we used activities in agreement with the linear response of the probes obtained with the count rate capability test.
The results of sensitivity and shielding tests were reported as count rate per unit of radioactivity (cps/MBq). The results of spatial resolution were reported as millimetre (mm). Count rate capability values were reported as megabecquerel (MBq).

Shift Index
For simultaneous 125 I-seed and 99m Tc breast cancer surgery, probes have to detect and discriminate the signals from both radionuclides. Two unfavourable clinical situations were simulated in vitro: a 125 I-seed implanted below a 99m Tc injection site and a 125 Iseed implanted below a 99m Tc sentinel node. For both configurations, we considered the 99m Tc source at 10 mm depth and the 125 I-seed at 20 and 30 mm depth (Fig. 3). We compared the spatial resolution profiles measured at these depths in the low energy Fig. 3 In vitro simulations of two unfavourable clinical configurations window for both sources ( 125 I signal vs 99m Tc scatter signal). The count rates were multiplied by clinical activities without taking into account probes linearity issues: 1, 5 and 8 MBq for the sentinel node, the 125 I-seed and the injection site, respectively. When the 99m Tc scatter signal was higher than the 125 I signal, we defined a Shift Index as the minimum distance along the X axis needed to discriminate the two sources. The Shift Index was calculated as the limit of the "visible" FWHM of the 125 I signal, while moving the sources apart from each other (Fig. 4).

99m Tc scatter fraction
The 99m Tc scatter fraction, which only involved 99m Tc sources, is of major interest for the manufacturers. Indeed, they use to rely on its assessment to improve the structure of the crystal detector and the electronic system to reach the lowest scatter fraction achievable.
Compton interactions in the scatter medium or in the detection system, and electronic noise, are at the origin of the 99m Tc scatter signal. The scatter fraction index (SF) was defined as the ratio between the signal measured in the low energy window ( 99m Tc scatter) over the signal measured in the high energy window (photopeak).
Two SF were calculated, the first one as a function of activity (Activity SF) (Eq. 2) and the second one depending on the spatial positioning of the probe (Spatial SF) (Eq. 3) while performing the count rate capability and spatial resolution NEMA tests, respectively.

Results
The three probes worked properly and without any problem during the tests. The GammaSUP-II system needed a fifteen minutes warm-up before ensuring a stable measurement.  (Table 3) Sensitivity results are presented with their standard deviations but assuming a 0.5 mm spatial error positioning, a 15% error has to be added in quadrature [11]. The GammaSUP-II probe was more sensitive than the two other probes in air and scatter medium for both 99m Tc and 125 I sources. Regarding the SOE-3211 and SOE-3214 probes, the first one was more sensitive for the 99m Tc, whereas the second one was more sensitive for 125 I. For example, at 10 mm depth in water, the sensitivities for the SOE-3211, SOE-3214 and GammaSUP-II were, respectively, 2774, 2402 and 7865 cps/MBq with the 99m Tc source and 459, 1653 and 3106 cps/MBq with the 125 I-seed.
The sensitivity to scatter corrected from sensitivity through side shielding in air was equivalent for the 3 probes with the 99m Tc source (between 10.2 and 13.4 cps/MBq). With the 125 I-seed, the SOE-3211 probe showed better results (0.8 cps/MBq), while the SOE-3214 and GammaSUP-II probes seemed more sensitive to the scatter with 4.9 and 6.5 cps/MBq, respectively.

Spatial resolution in a scatter medium (Table 4, Fig. 5)
With the 99m Tc source, the GammaSUP-II probe allowed a better spatial resolution, while the SOE probes showed equivalent results (except for the FWHM at 10 mm depth). With the 125 I-seed, the SOE-3211 probe allowed a better spatial resolution while the SOE-3214 and the GammaSUP-II probe had equivalent results. For example, at 20 mm depth, the FWHM for the SOE-3211, SOE-3214 and GammaSUP-II probes were respectively 28, 29 and 25 mm with the 99m Tc source and 30, 34 and 34 mm with the 125 I-seed.

Side and back shielding
The side and back shielding of the three probes seemed to be efficient with results upper than 98% for 99m Tc and equal to 100% for 125 I.

Extended NEMA testing
Shift index (Table 5, Fig. 7) Injection site simulation: For the three probes, the 99m Tc scatter signal at 10 mm depth was higher than the 125 I-seed signal at 20 and 30 mm depth. The SOE-3214 probe allowed a better discrimination of the two sources. This probe led to the smallest  Sentinel node simulation: The 125 I-seed signal at 20 mm depth was higher than the 99m Tc scatter signal at 10 mm depth for the three probes, so no Shift Index was calculated. The 99m Tc scatter signal became higher than 125 I-seed signal at 30 mm depth for the SOE-3211 and GammaSUP-II probes, with a smaller Shift Index value for the SOE-3211 probe (30 vs 35 mm).
99m Tc scatter fraction (Fig. 8) Activity 99m Tc scatter fraction: The activity SF calculated from 0-1 MBq to 6.5-7.5 MBq according to the probes started from 10, 13 and 30% then decreased to 8, 10 and 12% for the SOE-3211, SOE-3214 and GammaSUP-II probes, respectively. The activity Fig. 6 Count rate capability of the three probes for 99m Tc and 125 I sources at 10-mm depth SF was linear over the whole activity range for the SOE probes and only above 5 MBq for the GammaSUP-II probe.
Spatial 99m Tc scatter fraction: For the SOE-3211 and SOE-3214 probes, the spatial SF calculated at every depth was around 10% at X 0 mm and increased to 20 and 30% respectively at X 50 mm . The spatial SF was higher for the GammaSUP-II probe, from 10% at (X 0 mm ; Z 10 mm ) to 70% at (X 50 mm ; Z 30 mm ).

Discussion
Studies reporting in vivo and in vitro (based on NEMA NU3-2004 standards) probes comparisons already exist in the literature and helped us to interpret and enforce the different tests [12]. They concerned 99m Tc sentinel node [13] and 18 F-guided surgery [14]. To our knowledge, such a characterisation had never been performed for 125 I-sealed sources. The impact of 99m Tc injections combined with intra-tumoural 125 I-seed-guided surgery has already been addressed through both in vitro and in vivo studies: I-seed at the same position at 3 cm depth) using lean chicken breast [16]. to overcome the 99m Tc downscatter [17].
In every study, a non-significant impact of 99m Tc scatter signal was demonstrated even for the Radioguided Occult Lesion Localisation (ROLL) technique with a low activity 125 I-seed (until 1.85 MBq).
The originality of our study was to extend the NEMA testing to calculate the Shift Index and 99m Tc scatter fractions. These indexes are two additional figures of merit to assess the probes performances for simultaneous nuclides detection, from the same NEMA protocol. In addition, the results from the spatial resolution test allow exhaustive in vitro simulations of any clinical configuration whatever the activity of the radioactive sources.

Standard NEMA testing
Before performing the NEMA testing, it was essential to begin with a repeatability test to assess the warm up delay necessary to ensure a stable measurement. The count rate capability test was the first NEMA test performed in order to define the activity range involving a linear response for each measurement channel.

Sensitivity
The sensitivity is related to the number of photons interacting into the crystal in the selected energy window. It depends on the material (structure, density) and volume of the detection crystal as well as on the material and thickness of the entrance layer. The GammaSUP-II probe was significantly more sensitive than the SOE probes for both 99m Tc and 125 I sources despite a reduced high energy window width. The best explanation comes from the volume of its 269 mm 3 crystal, about four times larger than the crystal of the two other probes.
The sensitivity difference between the two SOE probes with the 99m Tc source is also correlated with the crystal volume. Indeed, the 69 mm 3 cubic SOE-3211 crystal is larger than the 57 mm 3 orthocylindrical SOE-3214 crystal. Concerning 125 I sources, the stainless-steel entrance layer of the SOE-3211 probe absorbed a part of the low energy photons, implying a loss of sensitivity compared to the SOE-3214 probe.

Spatial resolution in a scatter medium
The spatial resolution depends of the solid angle of detection (crystal entrance surface, probe positioning) and the discrimination of scatter signal (energy window, entrance layer) for a better focusing efficiency.
With the 99m Tc source, the GammaSUP-II probe allowed a better spatial resolution despite a larger entrance surface thanks to its reduced high energy window width. With the same high energy window (80-170 keV) and same crystal entrance surface (Ø 5 mm), the SOE probes showed logically the same spatial resolution.
The 125 I source emitted low energy photons. Then a part of the signal, including scatter signal, was absorbed by the stainless-steel entrance layer for the SOE-3211 and GammaSUP-II probes, which improved their focusing efficiency so their spatial resolution. This phenomenon explains the fact the SOE-3211 probe had a better spatial resolution than the SOE-3214 probe despite the same crystal entrance surface (Ø 5 mm) and the fact the GammaSUP-II probe had the same spatial resolution as the SOE-3214 probes despite a larger crystal entrance surface (Ø 7 mm).

Shift Index
The Shift Index was calculated from in vitro simulations of clinical configurations. Even if these results cannot be directly transposed into a clinical application ( 99m Tc source volume, probe oriented by a surgeon), the methodology is of interest with the definition of a new figure of merit combining spatial resolution and signal intensity for an objective assessment of signals discrimination in such a context of simultaneous sources detection.
The Shift Index results highlighted a negligible impact of the 99m Tc sentinel node for the 125 I-seed detection but a potential lack of signals discrimination for a 125 I-seed implanted below a 99m Tc injection site. This can be correlated to our clinical experience, indeed the only case in which the surgeon had a difficulty to find the 125 I-seed concerned a lesion located under a 99m Tc injection site. Since then, it was decided to avoid 99m Tc injections in the breast quadrant where the 125 I-seed is implanted.

99m Tc scatter fraction
For the SOE probes, the activity SF was linear over the whole activity range, meaning the same saturation threshold and the same behaviour on both low and high energy windows. For the GammaSUP-II probe, the saturation threshold occurred at a higher activity in the low energy window explaining a higher and non linear activity SF below 5 MBq.
Due to attenuation and solid angle, the 99m Tc photopeak signal decreased according to depth (Table 3) and distance from the X axis origin (Fig. 5). The 99m Tc scatter signal, originating from both scatter medium and detector material also decreased with X and depth (validating our clinical simulations as the reflection of the worst detection geometry with the iodine seed aligned below the 99m Tc source). The spatial SF increased according to X for the three probes, meaning a higher relative contribution of scatter signal. However, the variation of spatial SF with depth was probe dependent (Fig. 8).
This methodology based on NEMA testing could be used by the manufacturers to improve the crystal structure and the design of the probes. In our experiments, the detector material and energy window chosen for the SOE probes allowed a reduction of the 99m Tc scatter fraction at lower activities and higher distances between the source and the probe.

Conclusion
No probe stood out from the others in the whole test series; however, several observations should be pointed out: -The three probes are suitable for distinct or simultaneous 125 I-seed and 99m Tc breast cancer surgery.
-The GammaSUP-II probe allowed a better sensitivity due to its crystal volume and better spatial resolution features with the 99m Tc source but saturated at a lower activity.
-Considering the 125 I-seed, the GammaSUP-II probe was more sensitive but the SOE-3211 probe showed the best spatial resolution.
-Concerning the Shift Index, the SOE-3214 probe allowed a better spatial discrimination for a 125 I-seed implanted below a 99m Tc injection site or sentinel node.
-The SOE probes were more efficient while performing the 99m Tc scatter fraction tests.
-The SOE-3211 probe, with overall polyvalent performances, seemed to be an interesting trade-off for simultaneous detection of 125 I and 99m Tc.
There is no perfect probe; the choice of the detection device has to be based on the best trade-off regarding the expected use. This in vitro study provided a complete and innovative characterisation and comparison of three different probes showing their own area of performance. Our approach based of NEMA standards allows objective and reproducible results.