ORIGINAL ARTICLE 
Annals of Nuclear Medicine Vol. 12, No. 1, 7-14, 1998 
Application of a beta microprobe for quantification of regional cerebral blood flow with 15O-water and PET in rhesus monkeys 
Chie SEKL Hiroyuki OKADA, Shinsuke MORI, Takeharu KAKIUCHI, Etsuji YOSHIKAWA, Shingo NISHIYAMA, Hideo TSUKADA and Takaji YAMASHITA 
Central Research Laboratory, Hamamatsu Photonics K.K. 
A beta microprobe was successfully applied to monitor arterial input function for quantification of regional cerebral blood flow (rCBF) in the monkey brain with 150-water and positron emission tomography (PET). The sensitivity of the probe was approximately 0.83 to l.67 cps/kBq/ml depending on the studies. A preliminary study was performed to find a suitable use and to evaluate the performance of the system and data analysis procedure. The results showed that dispersion correction of measured input function was unnecessary if microprobes were connected directly to the arterial catheter. Then multiple CBF measurements were done in three monkeys under anesthesia. Identical regions of interest were placed with the aid of magnetic resonance imaging (MRI) of each monkey and rCBF values were estimated. Estimated rCBFs were reproducible for several measurements. The mean CBF value for a pentobarbital anesthetized monkey was 46.0 ml/min/100 g (PaCO2 = 46.3 mmHg). This shows that the use of the beta microprobe for quantification of rCBF with PET was validated. The lack of a need for dispersion correction of observed input function is an advantage with the beta microprobe system because the probes are small enough to be placed near the arterial sampling site. 
Key words: regional cerebral blood flow, positron emission tomography, arterial input function 

INTRODUCTION 
IT IS IMPORTANT to measure absolute regional cerebral blood flow (rCBF) in both basic and clinical studies not only for the evaluation of rCBF but for the quantification of regional oxygen extraction and consumption. Positron emission tomography (PET) is widely used for noninvasive measurement of rCBF. 150-water and 15O-butanol are used as tracers for the rCBF measurements. Since these tracers are usually administrated as an intravenous bolus injection, one of the difficulties in the quantification of rCBF with PET is the accurate measurement of arterial input function. The arteries are too small and radioactivity in the arteries changes too rapidly for the PET system to detect it. Manual sampling of arterial blood from 
Received June 5, 1997, revision accepted October 1, 1997. For reprint contact: Hiroyuki Okada, Central Research Laboratory, Hamamatsu Photonics K.K., 5000 Hirakuchi, Hamakita, Shizuoka 431-8601 , JAPAN. E-mail: okada@crl,hpk.co.jp 
peripheral arteries and beta detector systems, have been applied to monitor the input function of the tracer into the brain. Generally measured input function is smeared out by a significant volume of blood in the detector and the passage (e.g. tubes) between the sampling site and the detector. Because direct use of the dispersed input function generally induces overestimation of rCBF, mathematical corrections are necessary with those methods.1-3 We have developed a fiber-type beta-microprobe system for the direct detection of local beta particles.4,5 In this study, we applied the beta-microprobe system for the measurement of arterial input function of 15O-water during the PET scan. Because the microprobes are small, they were able to be positioned close to the arterial catheter to minimize possible dispersion of measured input function. First, we attempted to find a suitable method for the quantification of rCBF with PET and the beta microprobe in a monkey. Eight PET and microprobe measurements were performed and a microsphere method was applied to validate the method. Then a total of 17 measurements in three anesthetized monkeys were done to evaluate the reproducibility of the rCBF values obtained with the PET and beta microprobe system. 

MATERIALS AND METHODS 
Beta Microprobe System The beta microprobe system was developed especially for measurement of positron-emitting tracers in small animals because the fiber-type probes can be inserted directly into the target organs. Since the effective volume is up to the positron range of the isotopes, it is smaller than spatial resolution of PET. Furthermore, the combination of small probes and optical fiber facilitates the positioning of the probes. As Figure 1 shows, the beta microprobe system consists of plastic scintillation fiber probes, optical fibers, photomultiplier tubes, a signal amplifier, discriminator and counter. As the scintillation light diminishes to a single photon as it passes through the optical fiber, the photomultiplier tube for photon counting (R2295, Hamamatsu Photonics, K.K., Japan) was chosen. Output signals are stored in a PC-9801 personal computer (NEC, Japan). The shortest sampling time with this system is one second, and the optimal sampling time can be fixed for the purpose. The computer acquires counts during the fixed sampling time. Up to 6000 data points can be stored. The time activity curves of the probes are displayed simultaneously during the measurement. The scintillation fibers are 0.5 mm in diameter and 3.0 mm in length. To measure b+-rays, a pair of probes are needed as shown in Figure 2. The measurement (M)-probe detects both b+-rays and background gamma-rays while the other, the reference (R)-probe detects only Y-rays. 
Calibration of the Beta Microprobe 
The linearity of the beta microprobe was evaluated up to 105 cps (approximately 100 MBq/ml) prior to the following experiments (unpublished data). Radioactive counts generated purely by b+-rays, Qb [cps], were obtained by weighted subtraction of the counts of the R-probe, QR [cps], from those of the M-probe, QM [cps]: 
Where r is the ratio of sensitivity for the gamma-ray of the two probes. The r was obtained under the condition in which only gamma-rays were measured. The activity concentration of b+ particles, Cb [kBq/ml] is relative to Qh: 
where CCF is the cross-calibration factor for the probes [cps/kBq/ml]. From equations (1) and (2), CCF was obtained with a known activity concentration Cb [kBq/ml] : 
In vitro measurement was performed to obtain CCF. The ratio of sensitivity for gamma-rays of M- and R-probes was determined from count rate of the attached natrium line source. Then microprobes were immersed in 150-water and data were acquired for 10 seconds. At the beginning of the measurement, 1 ml of 150-water was sampled and counted in an autowell gamma counter (AUTO WELL GAMMA SYSTEM ARC-2000, Aloca, Japan) which had already been cross-calibrated with a dose calibrator (CURIEMETER 3, Aloca, Japan). All data were corrected for the physical decay of 150 to the starting time. CCF was then calculated with equation (3) by using the averaged count rate. CCF from the in vitro measurement above was applied for the preliminary study to estimate rCBF. After the preliminary study, numerous measurements in different animals were performed and the necessity of individual CCFs for individual animals was found. Therefore in the rest of the measurements, individual CCFs were calculated, and for each PET scan an arterial blood sample was withdrawn during PET and microprobe measurement. The sample was weighted and counted in an autowell gamma counter. The CCF value was obtained from the activity concentration of the sample and beta microprobe outputs at the time when the sample was withdrawn. The calibration was performed in every PET study and the CCF value was determined by averaging the values obtained in repeated measurements in individual monkeys. 
I. Preliminary Study 
Preliminary experiments were performed in a monkey. A total of eight PET and microprobe measurements were done under several conditions of microprobe measurement, such as different rates of withdrawal of arterial blood and different distances from the arterial catheter to the probe, to investigate observed input functions. To evaluate estimated rCBF, we attempted to compare rCBF values estimated with the combination of PET and the beta microprobe and those with the microsphere method. The physiological conditions of the monkey were maintained throughout the experiments. 
Animal preparations 
Monkeys were maintained and handled in accordance with recommendations by the US National Institutes of Health and also the guidelines of the Central Research 
Laboratory, Hamamatsu Photonics K.K. One male rhesus monkey (5.5 kg) was used in the preliminary study. The monkey was anesthetized with intravenous pentobarbital (25 mg/kg). Maintenance doses of anesthesia were administered as needed. A catheter was placed in a femoral artery for measurement of input function with beta microprobes and another was placed in a saphena for administration of 15O-water and return of the arterial blood. Arterial blood pressure, body temperature and heart rate were monitored throughout the study. Arterial partial pressure of O2 (PaO2), CO2 (PaCO2) and pH were measured with a blood gas tension analyzer (NOVA Stat Profile 3 Analyzer, NOVA Biochemical, USA). Two pairs of beta microprobes were connected to the arterial line. The microprobes were set as described in Figure 3. Each pair of microprobes were encased in lead to minimize background gamma-rays. Arterial blood was with-drawn with a peristatic pump (PERISTA BIO-MINIPUMP, AC-21 20, ATTO, Japan) at a constant rate (5 ml/min or 10 ml/min) and was returned to the saphena to avoid excessive loss of blood. When manual samplings were applied, arterial blood samples were withdrawn from a three-way cock which was connected between the probes and the extension tube. The probes and the tube were flushed with heparinized saline after every PET measurement. 
Conditions for beta microprobe measurement 
(1) Distance from arterial sampling site Two pairs of microprobes were used in this study to investigate the influence of the path length and dead volume of the tube between the probes and the catheterized femoral artery. A pair of microprobes was placed directly into the catheter in the artery and another pair was connected to it by way of a manometer tube of 50 cm long and 0.5 mm in inner diameter. The volume of the tube was 0.4 ml. Output data for the probes were acquired simultaneously and compared so that two input functions were measured simultaneously in one PET measurement. (2) Rate of withdrawal of arterial blood The rate of withdrawal of arterial blood for monitoring input function was considered to affect observed input functions. Arterial blood was withdrawn at the constant rate of 5 ml/min or 10 ml/min with a peristatic pump. 
Manual samplings 
Manual arterial blood samplings were applied during PET scan and beta microprobe measurement in a study to compare measured input function detected with the beta microprobe and that with manual sampling. The arterial samples were taken approximately every five seconds and the sampling times were recorded. The samples were weighed and the radioactivity in each sample was counted in an autowell gamma counter. Then the input function was constructed with the sampling times and the radioactivity concentrations of the samples. Beta microprobe measurement Beta microprobe data acquisition lasted for 150 seconds after the tracer injection. At 120 seconds after the beginning of the measurement, approximately 1 ml of arterial blood was withdrawn. The sample was weighed and the radioactivity was counted in an autowell gamma counter, which had already been calibrated with a dose calibrator. The ratio of gamma sensitivity of two microprobes was evaluated by microprobe measurement after the injection of 150-water without withdrawal of arterial blood from each monkey. 
PET scan protocol 
The monkey was placed in a PET scanner (SHR2400, Hamamatsu Photonics K.K., Japan) in the supine position. Details of the PET system are given elsewhere.5 The scanner consists of five detector rings covering a field of view (FOV) 291 mm in diameter by 74 mm in the axial direction. The scanner simultaneously acquires 9 planes with an interplane spacing of 6.75 mm. The transaxial resolution of the scanner was 2.8 mm full width at half maximum (FWHM) measured with an 18F line source (inner diameter of 1.0 mm) in the center of the FOV. That of 15O-water images shown in Figure 4 was 4.5 mm FWHM. The axial resolution was 5.7 mm FWHM in the direct plane and 5.3 mm FWHM in the cross plane. The head of the monkey was fixed in a head holder developed by Hamamatsu Photonics and positioned in the scanner parallel to orbitomeatal (OM) line in the second slice of the PET image. Just after a bolus injection of 1.2 GBq of 15O-water in 1.5 ml saline, PET scan and microprobe measurement were started simultaneously. Twelve frames of 10 second PET data were acquired dynamically. Altogether eight PET measurements with different peristatic pump flow rates and with or without manual samplings: three withdrawal rates of 5 ml/min without manual sampling, three 10 ml/min without manual samplings and two 5 ml/min with manual samplings, were performed in the monkey with interval times of 10 minutes to allow the radioactivity of O-15 to decay. Figure 3 shows the outline of the settings. 
Microsphere method 
The standard microsphere method was applied to evaluate rCBF with PET and the beta microprobe system. 99mTc labeled macroaggregated human albumin particles (99mTc-MAA, Dai-ichi Radiopharmaceutical Laboratory Inc., Japan) were used for this purpose. Ninety percent of the particles are 10 to 60 um in diameter and none is larger than 100 um so that all of them should be captured in arterioles or capillaries. After a series of PET measurements, a left thoracotomy was performed and 0.48 GBq of 99mTc-MAA in 1 ml saline was injected directly into the left ventricle. Arterial blood was collected with a Harvard pump at the rate of 5 ml/min for 5 minutes from the left femoral artery. Immediately after the procedure. The animal was killed with an overdose of pentobarbital. The brain was removed and sliced as PET image planes. rCBF by the microsphere method was calculated as follows: 
where F is the rate of withdrawal of arterial blood [ml/ min], Cm is the tissue microsphere activity concentration [kBq/g] and Qmb is the microsphere radioactivity of the arterial blood [kBq]. rCBF measured by the microsphere method was used as the reference rCBF. 
Regions of Interest 
When brain tissue samples were taken from the slices, photographs of the slices were taken in order to record the locations of the tissue samples. Then regions of interest were placed on the PET images in identical to those locations of the samples. 
Estimation of rCBF 
PET and microprobe data were decay corrected to the time when each measurement was started. PET data and beta microprobe data were calibrated in units of kBq/ml. The density of the brain tissue was assumed to be 1.04 g/ml. Kety's model7 was applied for the tracer kinetic model of water in the brain. Integration of the differential equation and non-linear least squares minimization were applied to estimate rCBF and other parameters.8,9 Estimated parameters were uptake rate Kl (= FE) [ml/min/g], washout rate k2 (= FE/p) [min-1], and arrival time shift of the tracer between brain ROI and beta microprobes delta-t [min]. Dispersion of observed input function with beta microprobes was also estimated in the preliminary study.3 
where Ci(t) is the brain tissue activity concentration of i-th ROI [kBq/g]. F is rCBF, E is the fraction of water extracted from blood to tissue, At is the tracer arrival time lag between the ROI and beta microprobe, p is the blood brain partition coefficient of water [g/ml], Ca(t) is the true input function [kBq/ml], I(t) is the measured input function [kBq/ml] and t is the dispersion constant [min]. In this study, it was assumed that the brain tissue region observed with PET was homogeneous and that E was 1. In addition, the partial volume effect of PET was ignored. 
Evaluation of estimated rCBF 
There were two measured input functions in each PET study. One was measured with microprobes connected directly to the arterial catheter and the other was connected via a tube. Two CBF estimation methods, with and without dispersion correction, were applied. To evaluate the accuracy of the estimated rCBF, we calculate R as the ratio of estimated rCBF to reference rCBF: If R is 1 , it means that the estimated rCBF agrees with the reference rCBF. Then the mean ratios of the same conditions such as the pump flow rate, position of the probes and with/without dispersion correction were calculated. 
II. Main Studies 
Three male rhesus monkeys (5.35 +- 0.33 kg) were used. Protocols of animal preparations and PET scan were the same as for the preliminary study. The microsphere method was not used. Beta microprobes were connected directly to the arterial catheter. Five or six of PET and beta microprobe measurements were done in each monkey. The total number of measurements was 17. Beta microprobes were connected directly to the arterial catheter. The rate of withdrawal of arterial blood was 5 ml/min or 10 ml/min with a peristatic pump. In order to choose identical regions of interest in the monkeys, brain MRI images of the monkeys were obtained prior to the PET studies. 10 regions of interest (ROIs) were placed on MRI image slices of each monkey, and matched with PET image planes. The regions of interest are shown in Figure 4. Since it had been found that dispersion of input function can be ignored if it is monitored close to the arterial catheter, rCBF, partition coefficient and delay were estimated by a non-linear least squares method from arterial input function measured with the beta microprobe and brain time activity curves measured with PET. 

RESULTS 
Physiological Conditions Mean PaCO2 was 46.3 +- 1.39 mmHg and mean arterial systolic and diastolic blood pressures were 139.3 +- 11.5 and 78.3 +- 15.8 mmHg. Mean heart rate was 130 +- 13. Body temperature was maintained at 37deg.C with a heating pad. 
Cross Calibration Factor for Beta Microprobe 
The ratio of sensitivity for gamma-ray of M- and R-probes was 0.98. The radioactivity concentration of the water was 6.18 MBq/ml and the calculated CCF was 0.781 cps/kBq/ml. When CCFs were obtained in the monkeys during the PET studies, they differed from monkey to monkey and varied from 0.853 to 1.536 cps/kBq/ml. 
I. Preliminary Study 
(1) Observed input functions 
Arterial input functions obtained under various conditions are shown in Figure 5. Significant dispersion of measured input function was observed with beta-microprobes connected through a tube. Input function obtained with manual samplings was not frequent enough to trace the rapid changes in the arterial tracer concentration. (2) Evaluation of estimated rCBF Table 1 shows the ratio of estimated rCBF and reference rCBF. The use of input function measured with probes connected via the tube without dispersion correction resulted in significant overestimation of rCBF, and a slower arterial blood flow rate caused overestimation of rCBF when input function was measured with probes via the tubes. On the other hand, the use of input function measured with probes connected directly to the arterial catheter resulted in close agreement between estimated rCBF values and reference rCBF values even without dispersion correction. And the rate of withdrawal of arterial blood did not affect the ratio. In addition, when dispersion correction was performed, the estimated T was almost 0 s for input function measured with the beta microprobe connected directly to the arterial catheter and about 5 s via the tube when dt (delay) was fixed. Dispersion correction was therefore necessary for input functions measured with microprobes connected via a tube but it was unnecessary when they were measured with probes connected directly to the arterial catheter. Although each estimated rCBF value was different from those obtained by the microsphere method, the mean estimated CBF value and mean microsphere CBF value were equal. Especially the mean CBF value estimated without dispersion correction on input function and measured with probes connected directly to the catheter agreed closely with the mean microsphere CBF. It is evident that dispersion correction is unnecessary if microprobes are set close enough to the catheter. 
II. Main Study 
Table 2 shows rCBF obtained from 17 measurements in three monkeys. There was no significant difference between estimated rCBF values for two different rates of withdrawal of arterial blood (5 and 10 ml/min). This agrees with the result of the preliminary study. The mean CBF value for the pentobarbital anesthetized monkey was 46.00 ml/min/1OO g when PaCO2 was 46.3 mmHg. 

DISCUSSION 
Dispersion correction Table 1 indicates that dispersion of the measured input function can be negligible if microprobes are connected closely enough to the arterial catheter. It is considered that the small size of the monkey lowered relative dispersion between arteries in the brain and microprobes connected to the catheter in the femoral artery. In fact there were slight differences between measured input functions in the femoral artery and the common carotid artery in a monkey (Fig. 6). This suggests that the longer passage from the heart to the femoral artery smeared the input function more than that of the common carotid artery. But the "true" input function for brain tissue was considered to be distorted further by additional pathways to the brain tissue, and became almost equivalent to the input function observed in the femoral artery. Furthermore, that there was no significant difference between estimated rCBF values for the two withdrawal rates (5 and 10 ml/min) indicates negligible dispersion of the measured input function with the beta microprobe. A faster rate of with-drawal of arterial blood is desirable to minimize dispersion of the measured input function if there is a significant dead volume between the arterial catheter and the probes. On the other hand, an excessive withdrawal rate can alternate the blood circulation of the subject, which may result in a change in rCBF. But the use of the microprobe system minimizes the dead volume so that there is no need for a faster withdrawal rate, and it does not require mathematical correction for dispersion. It can facilitate the estimation of rCBF without sacrificing the accuracy of the measurement of input function. 
Possible improvements and applications 
In order to monitor the input function more accurately, microprobes can be inserted through major arteries to the target organ to minimize dispersion and loss of blood. The reference probe (R-probe) may not be necessary because background subtraction without withdrawing blood is enough to correct background gamma-rays from surrounding objects such as the auto-injector and cardiac blood pool. Appropriate settings of the probe, which minimize background counts, improve the accuracy of the calibrated radioactivity concentration. The probes and connected optical fibers should be placed away from highly radioactive objects such as the injector and the subject's body, and the use of a lead shield is also advisable. In this study, the microprobe system was applied for the measurement of the arterial input function of 150-water. This method can be applied for other tracers such as 18F-fluoro-deoxyglucose (FDG) and 11C labeled tracers. Although they require a plasma input function, several blood samplings during the PET scan should be enough for the purpose. The microprobe is advantageous for the early phase of the scan, which requires very frequent blood samplings. The system can be used for directly monitoring positronemitting tracers in target organs. For example, inserting the beta microprobe into a cardiac chamber through a femoral artery is possible not only for monitoring input function but also for evaluating of cardiac function. This strategy can also be applied during surgery on a tumor labeled with positron-emitting tracers which bind specifically to tumor cells. Not only faster temporal resolution than PET but also its practicality can be an advantage in animal studies. The system can be easily applied to small laboratory animals, and it should accelerate the development of useful positron-emitting tracers for clinical use. 

REFERENCES 
1. Meter E. Simultaneous correction for tracer arrival delay and dispersion in CBF measurements by the H215O autoradiographic method and dynamic PET. J Nucl Med 30: 1069-1078, 1989. 
2. Iida H, Higano S, Tomura N, Shishido S, Kanno I. Miura S, et al. Evaluation of regional difference of tracer appearance time in cerebral tissue using 150-water and dynamic positron emission tomography. J Cereb Blood Flow Metab 8: 285-288, 1988. 
3. Iida H, Kanno I, Miura S. Murakami M, Takahashi K, Uemura K. Error analysis of quantitative cerebral blood flow measurement using H215O autoradiography and positron emission tomography, with respect to the dispersion of the input function. J Cereb Blood Flow Metab 6: 536-545, 1986. 
4. Okada H, Tsukada H, Shimizu K, Yamashita T, Inoue O, Kobayashi K. Development and assessment of beta microprobe system. KAKU IGAKU (Jpn J Nucl Med) 29: 823, 1992. 
5. Seki C, Okada H, Mori S, Yoshikawa E, Kakiuchi T, Ando I, et al. Quantification of regional cerebral blood flow with PET using beta microprobe system for monitoring arterial input function. KAKU IGAKU (Jpn J Nucl Med) 31 : 802, 1994. 
6. Yamashita T, Uchida H, Okada H, Kurono T, Takemori T, Watanabe M, et al. Development of a high resolution PET. IEEE Trans Nucl Sci 37: 704-710, 1990. 
7. Kety SS, Schmidt CE. The nitrous oxide method for the quantitative determination of cerebral blood flow in man: theory, procedure and normal values. J Clin Invest 27: 476-483, 1948. 
8. Press WH, Flannery BP, Teukosky SA, Vettering WT. Integration of ordinary differential equations. In Numerical Recipes in Pascal. Cambridge, Cambridge University Press, pp. 599-632, 1989. 
9. Press WH, Flannery BP. Teukosky SA. Vettering WT. Modeling of data. In Numerical Recipes in Pascal. Cambridge, Cambridge University Press, pp. 547-589, 1989.