[1] 12w Assessment of the short-lived non-pure positron-emitting nuclide 120 I for PET imaging

[1] 271w Purpose: The non-pure positron-emitting iodine isotope 120 I (T 1/2 =81 min) is a short-lived alternative to 124 I. 120 I has a positron abundance more than twice that of 124 I and a maximum positron energy of 4 MeV. This study was undertaken to evaluate and characterise the qualitative and quantitative PET imaging of 120 I. Methods: 120 I was produced via the 120 Te(p,n) reaction on highly enriched 120 Te. The measurements were done with the Siemens scanner HR+ and the 2D PET scanner GE PC4096+. A cylinder containing three cold inserts and a phantom resembling a human brain slice were used to evaluate half-life, positron abundance and background correction. To analyse the image resolution, a 1-mm tube placed in water was filled with 120 I and 18 F. Comparisons with 18 F, 124 I and 123 I (measured with SPECT) were made using the Hoffman 3D brain phantom. Results: The half-life of 81.1 min was reproduced by the PET measurements. The PET-based positron abundance ranged from 47.9% to 55.0%. The reconstructed image resolution found with the HR+ was 5.4 mm FWHM (12.3 mm FWTM), in contrast to 4.6 mm (8.6 mm) when using 18 F. Erroneous positive and negative numbers of radioactivity found in the cold inserts became nearly zero when the background of γ-coincidences was corrected for. Images of the Hoffman phantom were inferior to those obtained when 18 F or 124 I was applied but superior to the 123 I-SPECT images. Conclusion: Our data show that 120 I of high radionuclidic purity can be regarded as a suitable nuclide for the PET imaging of radioiodine-labelled pharmaceuticals.

[1] 194w After 99m Tc, the single photon-emitting 123 I is the most commonly used radionuclide in single-photon emission computed tomography (SPECT). It can be relatively easily introduced into many molecules of interest for clinical as well as research purposes. Although 123 I-labelled molecules are the most "natural" SPECT compounds, their biodistribution kinetics cannot be quantified in the same way as is possible with radiopharmaceuticals used in positron emission tomography (PET). In spite of considerable progress over the last decade, SPECT quantitation still suffers from lower sensitivity, inferior resolution and uncertainty in correction for absorbed and scattered radiation. Furthermore, the minimum frame interval for SPECT acquisition-even with a three headed camera-is too long to sample fast radiotracer kinetics adequately. Under these circumstances, an appropriate positron-emitting iodine isotope would be helpful for a PET-based complementary evaluation of radioiodine-labelled pharmaceuticals. For such an approach 124 I has been suggested and applied [1][2][3][4]. 124 I has a low positron abundance of 22% [5] and a long half-life of 4.18 days. Both factors lead to a high radiation dose, so that the use of 124 I is commonly limited to oncological applications, where the therapeutic radiation dose is intrinsically high.

[2] 58w We looked for a more favourable iodine isotope with a shorter half-life and higher positron abundance. If such an isotope is short-lived compared with 123 I, a doubleradioisotope procedure becomes possible: a radiopharmaceutical labelled partly with 123 I and partly with the positron-emitting iodine is injected. The fast dynamics directly after injection are recorded in the PET scanner.

[3] 12w Later, after the decay of the positron-emitting iodine, the SPECT acquisition follows.

[4] 198w Possible candidates for this objective were the iodine isotopes 119 I, 120 I [5-8], 121 I [6] and 122 I [9,10]. In spite of its favourable positron energy of 934 keV, 119 I must be excluded because of its half-life of only 20 min, which would require high amounts of radioactivity at the beginning of a radiosynthesis, as in the case of 11 C radiochemistry. It is, however, unrealistic to produce about 50 GBq of 119 I as a prerequisite for performing a synthesis. The same arguments hold for 122 I, with a half-life of 3.6 min. It is generally obtained from a 122 Xe-122 I generator [9,10], which is, however, sparsely available. 121 I has a convenient half-life of 125 min and a positron energy of just 509 keV, but it was not considered further because of the low positron abundance of only 13% [11], the necessity of using a medium-energy cyclotron for its production, and its decay into radioactive 121 Te. 120 I was regarded as a more relevant isotope. Its reported half-life is 81 min [11], as confirmed recently by Hohn et al. [12], who measured 81.5±2 min as the half-life of the 511-keV radiation.

[5] 216w The radionuclide 120 I was first characterised radiochemically by Butement and Qaim [6] via the 127 I(p,8n) 120 Xe ðβ þ ; ECÞ 120 I process in irradiations of NaI with 340-MeV protons. Whereas several γ-rays and two positron groups of maximum energies 2.1 and 4.0 MeV were identified, the exact positron branching was not known. Later studies reported positron abundance values between 39% [13] and 81% [11,14], and an isomeric state 120m I, with a half-life of 53 min, was also identified [14]. In an attempt to produce 120 I for medical use, Zweit et al. [7,8] performed some preliminary studies on the 122 Te(p,3n) reaction. The product was, however, a mixture of 120m I and 120 I, and thus not suitable for clean and quantitative studies. In a series of papers published from Juelich [12,[15][16][17][18], dealing with 122 Te(p,3n), 120 Te(p,n), and 120 Te(d,2n) reactions, it has been shown that the 120 Te(p,n) reaction over the proton energy of 16 to 9 MeV leads to high-purity 120 I, the level of 120m I impurity at end of bombardment (EOB) being about 4%. Using such a pure sample and studying the relative contributions of the annihilation radiation and the K X-rays of Te, the positron emission intensity of 120 I was determined as 56±3% [12].

[6] 124w Many of the positron emitters which do not emit positrons exclusively radiate single photons accepted by the energy window of the PET detector. Furthermore, in most cases a considerable percentage of the single photons is produced simultaneously with the positrons. If such a single photon and one of the annihilation photons hit the PET detectors within the coincidence window, the photons create a false coincidence-a so-called γ-coincidence. As the directions of the single and the annihilation photon are not correlated, the γ-coincidences result in a flat background similar to the background caused by random coincidences. PET systems are not designed to correct explicitly for the γ-coincidences, which affect the quali-tative and quantitative accuracy of PET imaging of nonpure positron emitters such as 120 I.

[7] 32w The aim of this study was to evaluate several parameters important for both qualitative and quantitative PET imaging with 120 I and to characterise the performance of PET measurements using this radionuclide.

[8] 86w As mentioned above, the γ-coincidences produced by the non-pure positron emitters result in a flat background. The impact of this background was studied with the three-rod phantom, which had been scanned with the HR+ scanner in 2D and 3D modes. These tests could be compared with the equivalent tests using the pure positron emitter 18 F. Sinograms were displayed with the MPItool and horizontal profiles averaged over 21 adjacent lines of response (LORs) were defined at identical locations which crossed the complete scanner's field of view.

[9] 169w The subtraction of the background was done as in a previous paper [28] and by other authors [4,26,29]. The background was estimated from the exterior edges of the sinograms. In each line (corresponding to one of the 144 projection angles), the 20 outmost bins on each side were averaged. The linear connection between the two means was regarded as background and subtracted from the measured coincidences. In the paper of Buchholz et al. [28], we found that for the HR+ scanner the background counts must not be subtracted completely. Therefore, a fraction of 75%, 80%, 85%, 90%, 95% or 100% of the estimated background was subtracted from the originally measured sinogram. The resulting sinograms (in 2D and 3D mode) were reconstructed as already detailed above. Here the filter width of the reconstruction Shepp filter was 4 mm. Reconstructed radioactivity concentrations were obtained from ROIs placed over the three rods (teflon, water and air) with a diameter of 75% of the rod diameter and within the area of hot radioactivity.

[10] 12w When these tests were done, the PC4096+ scanner was no longer available.

[11] 196w In order to obtain a realistic impression of the qualitative image performance of 120 I, a Hoffman 3D brain phantom was filled with 120 I, and in an equivalent study with 18 F. For attenuation correction a position matching 10-min transmission scan of the cold Hoffman phantom was utilised. The corrected sinograms were reconstructed as described above for the half-life test. For both 120 I and 18 F images, ROIs were drawn over cortical (GM) and WM areas (centrum semiovale) so that the GM/WM ratio could be calculated. For further comparison, the same procedure was carried out with 124 I and with 123 I, the classical iodine isotope for SPECT. The SPECT acquisition took 35 min with the triple-headed SPECT scanner Trionix TRIAD equipped with fan-beam collimators. The reconstruction parameters were those used for standard brain studies when an 123 I-labelled radiopharmaceutical such as 3-[ 123 I]iodo-α-methyl-L-tyrosine [30] is administered: filtered backprojection with a Butterworth filter (high cut 0.35, roll off 3.0), Chang's first-order attenuation correction with μ=0.1/cm and a contour finding procedure for each slice, no scatter correction. Using the MPItool, all image data sets were aligned to each other prior to the ROI definition.

[12] 35w In addition, the subtraction of 85% background as described in the last section was applied to the 120 I study of the Hoffman phantom to test the impact of this subtraction on the image quality.

[13] 67w When the test of the positron abundance was applied to the HR+ data after a background subtraction of 85% for the 3D scan and 75% for the 2D scan, the resulting positron abundance decreased to practically the same values for 3D and 2D: 41.6% and 41.7%. The relevance of these findings will be addressed in the Discussion in respect to the work of Kull et al. [29].

[14] 254w The comparative display (Fig. 5) of a transaxial section of the Hoffman 3D brain-phantom clearly indicates that the Fig. 4. Images of the three-rod phantom filled with 120 I without background subtraction (a), with subtraction of all background (b) and with subtraction of 85% of the background (c). On an imaginary clock face, the teflon rod is located at 12 o'clock, the air rod at 4 o'clock and the water rod 8 o'clock The percentage data given here were obtained by relating the radioactivity concentration in the ROIs over the teflon, water and air rods and in the exterior to the radioactivity concentration found in the 120 I-filled part of the phantom. The last line offers data obtained with the 18 F-filled phantom and without a background subtraction (BG) use of 120 I results in a loss of image resolution compared to equivalent measurements with 18 F. Also compared to 124 I, the 120 I image looks more blurred. When the GM/WM ratio was determined, it was 1.6 for 120 I, 2.2 for the 18 F and 1.8 for 124 I. Because the reconstruction procedures used for the 120 I and the 123 I images are different, it is difficult to compare the quality of the two images. Nevertheless, the 120 I image appears to be superior. The GM/WM ratio of the 123 I-image was 1.4. When 85% of the sinogram background was subtracted prior to reconstruction, the 120 I image showed an increased contrast and a greater GM/WM ratio of 1.9 (Fig. 6).

[1] 118w The resolution of the reconstructed image of the capillary 120 I source at FWHM was 5.4 mm; it was 4.6 mm when the line source was filled with 18 F. A greater difference was found for the FWTM, with values of 12.3 mm and 8.6 mm for 120 I and 18 F, respectively. As depicted in Fig. 1, the 120 I profiles are characterised by wings more extended than those of the 18 F profiles. One has to take into account that the line profile caused by the positron range itself is not Gaussian, but cusp-like in shape [31], so that the FWHM is less affected by the high positron energy of 120 I than is the FWTM.

[2] 155w Time-activity curves obtained from reconstructed but not decay-corrected images of the three-rod phantom scanned with the HR+ scanner yielded a half-life of 81.2 min for the 2D mode and 81.1 min for the 3D mode, respectively. When the half-life was determined from the directly measured true coincidences, values of 89.7 min (2D) and 94.4 min (3D) were found if all time points of the timecoincidence curves were considered. The corresponding mono-exponential fits were not satisfactory, indicating that these values for the half-life were not reliable. At the first two time points the dead time was 17% and 7% for the 2D mode and 24% and 11% for the 3D mode. At the last two time points the dead time was 3% and 1% for the 2D mode and 5% and 3% for the 3D mode. The half-life derived from just these two time points was 83.8 min for 2D and 86.4 min for 3D, respectively.

[3] 48w The PET reconstruction yielded the positron emissionbased radioactivity. When the reconstructed radioactivity of the GM chamber of the two-chamber phantom was related to the total radioactivity of 120 I measured by γ-ray spectroscopy in the three samples taken from the GM chamber, the resulting positron abundance was 50.2±4.9%.

[4] 238w When reconstructed images of the three-rod phantom scanned with the HR+ were evaluated as described above, a positron abundance of 47.9%±0.3% and of 55.0%±0.3% was found in the case of the 2D and 3D acquisition, respectively. The impact of this background on cold regions within an 120 I-filled phantom is documented by Figs. 3 and 4 and Table 1. In images reconstructed with the standard procedure from the unchanged 3D sinograms, the cold teflon insert showed about 68% of the radioactivity concentration of 2.73 kBq/ml found in the 120 I-filled compartment. In contrast, an ROI over the air-filled insert delivered a "negative" radioactivity concentration of -0.3 kBq/ml. Subtracting the whole background (Fig. 2) resulted in negative numbers in the teflon rod, but positive numbers in the air rod. When not all of the background was subtracted, it was possible to get radioactivity concentrations in all cold rods near to zero. The optimum amount of subtracted background was 85%. This optimum was defined as the minimum root of the summed squares of the percentage radioactivity in the cold rods. When an equivalent test of background subtraction was done for the 2D scans of the three-rod phantom, the amount of subtracted background for which the reconstructed radio-activity in the cold rods could be minimised was 75%. The radioactivity concentration measured outside the phantom was nearly zero (less than ±2%), independent of the amount of background subtraction and the acquisition mode.

[1] 158w This work deals with some issues related to PET imaging of the non-pure positron-emitting iodine isotope 120 I. Over recent years, several reports have been published on PET imaging using non-pure positron emitters, such as 124 I [4,27,32], 86 Y [28,33,34] or 76 Br [26]. A general problem of these isotopes is the additional single γ-radiation which is emitted mostly simultaneously with the pairs of 511-keV annihilation photons. In combination with one of the 511-keV annihilation photons a single γ-ray may cause a socalled γ-coincidence. According to the nuclear data published in the Table of Isotopes [14], about 75% of the positron emissions of 120 I are accompanied by one or more simultaneous single photons. Nearly all of those γemissions include a 560-keV photon, which is accepted within the energy window of the discriminator ranging from 350 to 650 keV, so that it may produce a γcoincidence with one annihilation photon if both photons hit the detector ring.

[2] 191w The directions of the single photons are not correlated with the annihilation photons. Therefore, the γ-coincidences lead to a uniform background which is not related to the measured radioactivity distribution and is responsible for both qualitative and quantitative errors. The probability of the occurrence of γ-coincidences depends on the acquisition mode. Compared with 3D PET, the septa extended in 2D PET restrict the acceptance angle and in this way also the probability of γ-coincidences to be recorded. The PC4096+ PET scanner used in the test on positron abundance has strong 3-mm lead septa of 20 cm radial extension. Therefore, this scanner is less influenced by γ-coincidences than the other more recent scanners acquiring in the 2D mode. Some previous studies on 124 I [27] and 76 Br [26] utilised the PC4096+ PET scanner as well as the Siemens CTI scanner ECAT Exact HR+, which has smaller septa and a larger acceptance angle. They showed that quantitative errors with the PC4096+ PET scanner are much smaller than with the HR+. Therefore we can assume that the measurement of the positron abundance of 120 I done with the PC4096+ yielded more reliable data.

[3] 152w The values of the positron abundance for 120 I obtained with the HR+ in 2D and 3D are in the range of 50% and comparable with the value of 58.4±2.0% determined via γray spectroscopy alone [12]. The results of both PET and γray spectroscopic measurements are based on the use of practically pure 120 I, in contrast to previously reported data [7,8] where the 122 Te(p,3n) or other reactions gave rise to the additional positron-and γ-ray-emitting iodine isotopes 120m I and 121 I. It should, however, be pointed out that the value of the positron branching in 120 I determined via γ-ray spectroscopy is dependent on the assumed intensity of the 1,523-keV γ-ray (11.2%). An uncertainty in the intensity Fig. 6. Transverse slices obtained from the PET recordings of a Hoffman 3D brain phantom filled with 120 I (a) without and (b) with 85% background subtracted before the standard reconstruction was performed

[4] 74w of that γ-ray would thus lead to an uncertainty in the absolute activity of 120 I and consequently in the determined positron intensity (despite the very pure 120 I source). On the other hand, the positron branching determined via a comparison of the 511-keV annihilation radiation and the K X-rays led to an absolute value of 53±3% [12]. We therefore conclude that the PET measurement of 120 I reported here leads to quantitative results.

[5] 94w If one looks at the high background of γ-coincidences in the 3D sinograms shown in Fig. 2, one may expect a much higher positron abundance measured in this test and may be surprised about the agreement with the γ-ray spectroscopybased result. However, the scatter correction prior to the image reconstruction regards the additional background outside the phantom, which is due to the γ-coincidences, as being caused by scattering as well and aims also to correct for these "additional scattered coincidences". In this way the absolute quantitation of 120 I radioactivity is not so bad.

[6] 332w There are, however, considerable errors in cold areas which should be devoid of any measured radioactivity. The teflon rod has a high attenuation coefficient. Therefore, background counts present in the area of this rod are amplified by the attenuation correction so that the "radioactivity" in the teflon ROI amounts to 68% of the hot radioactivity in the cylinder. A similar situation holds for the water insert, with its lower attenuation coefficient and a measured "radioactivity" of 24%. In contrast, in the air insert a "negative radioactivity" of -11% is found, for which an overestimating scatter correction is probably responsible. If the background displayed in Fig. 2 is totally subtracted, the results are also not satisfying, as documented by the numbers in the second to last line of Table 1. A reason may be that the model-based 3D scatter correction as well as the convolution kernels applied in the 2D scatter correction expect some scatter to be corrected for, which has been removed by a complete background subtraction, too. In this situation we found that not all of the background must be subtracted to obtain radioactivity concentrations of about zero in all the three cold rods of the phantom. The amount of background to be subtracted is different in 3D and 2D, at 85% and 75%, respectively. For comparison, in a previous study [28] we found a value of 75% for both 3D and 2D when the phantom was filled with 86 Y, which has a different decay spectrum. In this work a linear fit was applied to the background. When Kull et al. applied a parabolic fit to the background found in a 2D study with 86 Y, their approach led to satisfyingly low radioactivity measured in the cold rods [29]. Thus, the final effect is similar to our incomplete subtraction of background. An appropriate background subtraction leads not only to reasonable results in cold areas, but also to improved image contrast, as demonstrated with the Hoffman 3D brain phantom (Fig. 6).

[7] 122w When the positron abundance was examined after having subtracted 85% (for 3D) or 75% (for 2D) background from the scan done with the HR+ scanner, the positron abundance became much lower, with values of about 42%. A similar decrease in the PET-measured positron radioactivity can be observed, for example, in Table 2 of the paper of Kull et al. [29]. However, when they applied their calibration procedure in addition to the background subtraction, the decrease was counterbalanced so that the finally obtained positron radioactivity increased again and differed only by -7% from the data without any background subtraction and specific calibration. Thus, it can be concluded that our test on the positron abundance without background subtraction and calibration leads to realistic numbers.

[8] 377w One major concern when examining 120 I is the high maximum positron energy of about 4 MeV [11]. Due to this high energy, the image resolution of 120 I is not optimal. The loss of resolution is better expressed by the parameter FWTM rather than FWHM. As already stated above, the positron range-related point spread function of a 120 I point source is not Gaussian but cusp shaped [30]. In this way a long positron range worsens the FWHM less than the FWTM. The positron energy yields not only an inferior image resolution, but also a decreased contrast, as seen in the 120 I images of the Hoffman 3D brain phantom. The comparisons between 18 F, 124 I and 120 I with regard to the visual impression of image resolution and to the contrast measured by the GM/WM ratio correspond to the increasing positron range from 18 F via 124 I to 120 I. On the other hand, a comparison with the SPECT image of the Hoffman phantom, for which 123 I was used, demonstrates that 120 I is adequate for comparative studies where the PET study would yield better temporal and quantitative data. As already mentioned above, one major motivation underlying the search for a short-lived positron emitting iodine isotope was the desire to obtain kinetic information about iodinelabelled radiopharmaceuticals, especially those with fast extraction, as a result of which short time frames must be acquired. Such information would not be available from the SPECT measurement alone. The idea is to co-inject the 123 I-labelled tracer with the 120 I-labelled one. The PET measurement is not influenced by 123 I. Rapid distribution kinetics of a PET tracer can be adequately recorded by frames with periods of, for example, 10 s, which is not achievable with SPECT even if a three-headed camera is applied. After the decay of 120 I, the SPECT measurements can be performed. Here, the advantage of the longer-lived SPECT isotope can be utilised to acquire dynamic data even many hours after injection. Furthermore, the PET data can help to calibrate the SPECT information. An equivalent approach has been suggested for the combined application of the positron-emitting 110m In (T 1/2 =69.1 min) and the single photon-emitting 111 In (T 1/2 =2.8 days) [35].

[9] 141w There are some other aspects relevant to the use of 120 I. Because of the higher positron energy and abundance of 120 I compared with 124 I, the radiation dose caused by a 120 Ilabelled radiopharmaceutical will be greater than that caused by the radiopharmaceutical labelled with 124 I, if the tracer's biological half-life is so short that the long physical half-life of 124 I does not play any role. In the case of a long biological half-life, the fast decay of 120 I compared with 124 I results in lower radiation of the 120 I-labelled radiopharmaceutical. Finally, it is well known that the specific activity which can be reached with radioiodine-labelled tracers is often higher than that obtained with the classical positron emitters 11 C and 18 F, which is especially important for the PET analysis of neurotransmitter receptor ligands.

[1] 77w This work presents several parameters important for both qualitative and quantitative PET imaging of 120 I and characterises the performance of PET measurements using this radionuclide. Previous literature data were based on 120 I of low radionuclidic purity. Our data show that pure 120 I produced using the 120 Te(p,n) reaction on highly enriched target material and a small cyclotron can be regarded as a suitable radionuclide for the PET quantitation of the biodistribution of radioiodine-labelled pharmaceuticals.

[1] 61w Production of 120 I 120 I was produced via the 120 Te(p,n) reaction over the energy range from 16 to 9 MeV [15]. Although the target material 120 Te is highly expensive, it was applied in order to obtain the minimum level of 120m I impurity. The optimised method of production and quality control have been described in detail elsewhere [5].

[2] 91w The total amount of radioactivity used for the tests described below was measured by γ-ray spectroscopy. For this purpose, a highresolution HPGe detector was used. It was coupled to an ORTEC (Spectrum ACE) 4K MCA plug-in card, which was connected to an IBM-compatible PC-AT. The detector counting efficiency was determined using standard sources. The total 120 I activity in each sample was obtained relative to the 1,523-keV γ-ray, whose absolute intensity is known to be 11.2% [14]. This γ-ray is not emitted by the small amount (~4%) of 120m I impurity.

[3] 83w One of the studies presented here was done with the GE/Scanditronix PET scanner PC4096+ [19], which is a 2D scanner with solid 3-mmthick lead septa of a radial extension of 20 cm. Therefore, the amounts of random as well as scattered coincidences were rather low as compared to other 2D scanners with smaller septa. With the PC4096+ scanner the random coincidences to be corrected were estimated using the measured single counts of each detector. For scatter correction the convolution-subtraction method [20] was employed.

[4] 119w The other experiments utilised the Siemens/CTI scanner ECAT Exact HR+ [21], which acquires emission data in the 2D or 3D mode. Most of the HR+ measurements reported here were obtained in the 3D mode, because it is more relevant for the future, when most or possibly all PET and PET/CT scanners will operate without septa, just in the 3D mode. The septa of the HR+ used in the 2D mode are made of lead, 0.8 mm thick and 8 cm long in the radial direction. Random coincidences were corrected by the delayed window technique [22]. In 2D scatter correction was based on the convolution-subtraction method [20], whereas the model-guided 3D scatter correction was done using the Klein-Nishina formula [23].

[5] 234w In order to examine the image resolution of the high-energy positron emitter 120 I within a scatter medium we filled capillaries of 1-mm inner diameter with 120 I and inserted them in a water filled cylinder of 20-cm diameter and 10-cm length. The test was recorded with the HR+ scanner operating in the 3D mode. In the NU1-1994 PET performance test [24] such an arrangement is used for the scatter test. With this test, scattering conditions similar to those found in real brain studies could be simulated. Here, the emission scan lasted for 15 min. The 10-min transmission measurement was done after the decay of the radionuclide. The acquired data were reconstructed with 3D filtered backprojection [25] using a Ramp filter with a cut-off frequency of 0.5 cycles/pixel. Prior to the reconstruction the data were corrected for attenuation and scatter. The pixel size of the reconstructed images was 0.5×0.5 mm 2 . The same procedure was repeated with 18 F filled in the line sources. In reconstructed transverse images the line sources were seen as points. Horizontal and vertical line profiles were defined crossing the maximum of the reconstructed point images. The reconstructed image resolution expressed as full-width at half-maximum (FWHM) was determined by linear interpolation between adjacent pixels at half the maximum pixel value. Correspondingly, the full-width at tenth-maximum (FWTM) was obtained. The reported results are averaged values over horizontal and vertical profiles.

[6] 85w The half-life of 120 I was measured in this study to check whether PET measurement is able to reproduce the literature value of 81 min [11]. A discrepancy might be caused by problems in the dead-time correction, as has been indicated by Lubberink et al. [26], who examined the non-pure positron emitter 68 Br with both a PC4096+ and an HR+ scanner. Using the half-life test, PET was implicitly examined in respect to the linearity of quantitative measurements of different concentrations of 120 I radioactivity.

[7] 174w The HR+ scanner was used to scan a water-filled cylindrical phantom (called the three-rod phantom in the following) that has a diameter of 20 cm and three cold inserts filled with teflon, water and air. The cylinder was filled with about 20 kBq/ml of 120 I and measured over 7 h with four repetitive and alternating scans performed in 2D and 3D modes. The sinogram data were corrected as described above, FORE rebinned, and reconstructed by the routinely used 2D filtered backprojection (Shepp filter, filter width 2.5 mm). Regions of interest (ROIs) were defined on that part of the reconstructed images which included the 120 I radioactivity, and timeactivity curves were constructed for the 2D and the 3D study. Each curve consisted of 16 data points distributed over a period of 7 h and was fitted by a mono-exponential function, from which the half-life was obtained. In addition, the half-life was derived from the time course of the acquired random corrected coincidences, which had been recorded in the header information of the sinogram files.

[8] 70w In contrast to 124 I, whose positron abundance of 22% is well recognised so that quantitative tests can rely on and use this number, literature values for the positron abundance of 120 I range from 39% [13] to 81% [11,14]. A recent study [12], in which highly pure 120 I was used, found a positron abundance of 56±3%. In this situation we wanted to determine PET-based numbers on positron abundance.

[9] 197w One test was done with the PC4096+ scanner, whose outcome is affected only slightly by the γ-coincidences because of its strong interplane septa. For this purpose we analysed a dynamic scan with six frames of 10 min of a phantom (further referred to as the twochamber phantom) which consisted of two compartments, simulating the rough shape of the cerebral cortex [grey matter (GM)] and white matter (WM) at the level of the basal ganglia [27]. The phantom is a straight cylinder of 10-cm length so that measurements are independent of the location in the z direction (the scanner's longitudinal axis) if the phantom's cross-section is oriented transversely. The GM and WM chambers are shaped in such a way that ROIs can be placed without evoking a partial volume effect. The phantom was filled with about 10 kBq/ml of 120 I. The recorded data were reconstructed with the routinely used filtered backprojection (Hann filter, filter width 4.5 mm) including all corrections for absorbed, random and scattered counts as well as dead time. The positron-based radioactivity concentration was derived from ROIs placed in the GM chamber area of the reconstructed images and averaged from slice 4 to slice 12.

[10] 34w The concentration of the positron radioactivity was also obtained from the half-life experiment with the HR+ and the three-rod phantom. The reconstruction procedure and the placement of the ROIs have already been explained above.

[11] 113w Having taken into account the radioactive decay, for which the literature value of 81 min [11] was used, the ratio between the mean ROI radioactivity concentration and the total radioactivity concentration yielded the positron abundance of 120 I. When analysing the PC4096+ study, the total radioactivity concentration was measured from three samples of 0.5 ml each taken out of the GM chamber. In the case of the HR+ experiment, the total radioactivity concentration resulted from the ratio of the 120 I radioactivity filled into the phantom and the phantom's water volume of 4,433 ml. In both cases the measurement of the total 120 I radioactivity was done via γ-ray spectroscopy as described above.