- Email: [email protected]
Intravenous 99mTc HM-PAO radionuclide angiography as an adjunct to spect assessment of cerebral disease
Computerized Medical Imaging and Graphics, Vol. 14, No. 2, pp. 91-95, 1990
0895-6111/90 $3.00 + .00 Copyright © 1990 Pergamon Press plc
Printed in the USA. All rights reserved.
INTRAVENOUS 99mTCHM-PAO RADIONUCLIDE ANGIOGRAPHY AS AN ADJUNCT TO SPECT ASSESSMENT OF CEREBRAL DISEASE A. Bartolini, B. Gasparetto, U. Ruffinengo, R. Amore, A. Prima'cera and C. Albano Institute of Clinical Neurology, University of Genoa CNR-Center of Cerebral Neurophysiology, Genoa, Italy
(Received24 March 1989;Revised 10 August 1989) Abstract--An application of intravenous radionuclide angiography as an adjunct to SPECT with 99roTe HM-PAO is presented. From the angiographic sequence a two-dimensional parametric image is generated computing, pixel by pixel, the center of gravity (COG) of the local time curves. This index is independent of blood flow and reflects the relation between transmitted and extravasated activity. COG value, compared in two symmetrical regions analyzed for differences in flow by SPECT, may help to state whether a detected SPECT changed uptake is linearly related or not to the corresponding change of blood flow. All 23 ischemic patients taken into consideration had a similar backflux in symmetrical regions with conspicuous changed uptake. In 3 meningiomas and 2 arteriovenous angiomas it was possible to detect the occurrence of a changed backflux in the lesion suggesting a nonlinearity between blood flow and 99mTCHM-PAO uptake.
Key Words: SPECT, 99mTcHM-PAO, Radionuclide angiography, Cerebral ischemia, Cerebral blood flow metabolic activity. Therefore the statement that the spatial distribution of 99mTCHM-PAO uptake is linearly related to that of blood flow is true only on condition that possible local changes in the permeability and tissue parameters in brain lesions do not determine substantial additional changes in 99mTcHM-PAO retention. Steady state 99mTc HM-PAO uptake, Qr of a monitored cerebral region, may be described as the product of flow, F, and integrated arteriovenous differences:
INTRODUCTION
99m-Technetium hexamethylpropyleneamineoxime (99rnTc HM-PAO) has been reported to cross the intact blood-brain barrier and to distribute in the brain in proportion to regional blood flow. Moreover it is retained sufficiently long to allow single photon emission computed tomography (SPECT) with widely available rotating gamma camera systems (1-6). At present, the lack of conclusive data concerning the fate of 99mTcHM-PAO in the brain does not allow rCBF to be quantified. A simple qualitative analysis based on the expected symmetry of 99rnTc HM-PAO distribution is, however, possible. Although the reliability of 99mTcHM-PAO uptake as a qualitative steady state indicator of cerebral blood flow has been supported by the high incidence of decreased 99mTcHM- PAO uptake in the side of brain ischemia (7), the possibility that factors other than blood flow may influence the actual 99mTc HM-PAO uptake cannot be ruled out. Generally speaking, the steady state uptake of 99mTc HM-PAO depends, in addition to blood flow, upon blood-brain barrier (bbb) permeability and mean residence time of the tracer within tissue. This last factor is governed by tissue affinity for 99mTc H/VI-PAO, and possibly reflects cellular and chemical composition and
Qr=F yf(C2(t)-Cv(t))dt
(l)
(Ca(t) and Cv(t)= arterial and venous concentrations). Assuming that integrated arterial activity is similar in all regions of the same patient, any change in uptake independent from blood flow will be associated with a change in the venous concentration curve. Radionuclide angiography may be easily performed following bolus injection of 99myc HM-PAO prior to SPECT examination. Analysis of local curves may provide, although on a simple planar basis, a way to compare in any patient the time course of the activity in regions analyzed for differences of flow by SPECT. The following study illustrates an application of radionuclide angiography, based on computer generation of center of gravity (COG) images, as a useful adjunct to the SPECT study. Our aim was to discover whether this simple, inexpensive technique was suitable for detecting regional changes in backflux and to what extent a
Address correspondenceto: ProfessorAlfredoBartolini, Centro di Neurofisiologia Cerebrale, Clinica Neurologica dell'Universita', Via De Toni, 5 16132Genova, Italy. 91
92
Computerized Medical Imagingand Graphics
changed backflux could explain a changed SPECT uptake.
PATIENTS AND METHODS
Patients Fifteen control subjects and 31 patients with a conspicuously changed 99mTC HM-PAO uptake were selected. The 15 control subjects underwent 99mTCHMPAO SPECT as a screening procedure for minor symptoms; this proved negative. In all subjects clinical observation and CT examination allowed us to rule out major cerebral disease. Of the 31 patients with changed 99mWC HM-PAO SPECT uptake, 29 had a decreased uptake and 2 an increased uptake; 23 suffered from cerebral ischemia clinically presenting with a completed stroke dating from 5 to 30 days prior to SPECT examination, 2 patients had an arterio-venous (av) angioma, and 6 had an intracranial tumor (4 meningiomas, 1 glioma, 1 metastasis). Diagnoses were based on the usual workup, including CT in all patients, and cerebral angiography when diagnostically necessary. All patients with changed SPECT uptake had an asymmetrical distribution of the tracer on visual inspection of sequential radionuclide angiography images.
March-April/1990, Volume 14, Number2
was calculated about the injection time and up to 120 secs. The choice of this time was based on the observation, reported by Leonard et al. (6), that the brain activity following intravenous injection slowly changes only in this interval. An image representing the distribution of COG12o was generated by summing up the original 120 frames, having multiplied each by the corresponding time delay from injection. The picture obtained was then divided by the sum of the original 120 frames. Externally-monitored brain time-activity curve Q(t) results from the product of blood flow (F) and the convolution of I(t), input function, with R(t) residue function or local unit impulse response (8,9):
(3)
Q(t) =Ff2I(u)(t-u)du.
I(t) represents the time-concentration curve crossing the local inflow and R(t) corresponds to the function which would have been recorded on the tissue for unit impulsive administration at the cerebral inflow. Using the Laplace transform, the COG120 of the local cerebral intravenous curve can be broken down as (8,9): 1"120
1"120
COG120=J0 tQ(t)dt/J o Q(t)dt=
Methods Radionuclide angiography was performed by monitoring the passage through the brain of a bolus of about 15 mCi of 99mTCHM-PAO injected into the right brachial vein. Following injection, a sequence of 120 frames at 1 sec intervals was started, with the patients positioned in the supine position, facing the detector. SPECT was performed 10 minutes later, using a rotating gamma camera (GE 400A) connected on line to a computer (GE STAR). The examination was carded out by taking 64 views during 360 degree rotation about the long axis of the patient and accumulating about 5 million counts for each examination. Tomographic reconstruction was performed by filtered back-projection (Ramp Hanning filter, cut-off frequency = 0.5 cycles/ pixel) and attenuation correction (K= 0.07).
120
1"120
1"120
1"120
tli(t)dt/J o I(t)dt+ J o tR(t)/J o R(t)dt
(4)
COG12o= COG12o(I)+ COG12o(R).
(5)
and
Assuming I(t) to be similar in all regions of the same patient, the regional COG12o values may differ among themselves only in the component COG12o (R). But R(t) may be described as resulting from the sum of intravascular transmitted activity T(t) and extravasated activity E(t) (9):
R(t) = T(t) + E(t)
(6)
2°
(7)
Theoretical considerations Sequential radionuclide angiography flames were elaborated with a view to obtaining a two-dimensional representation of the center of gravity (COG12o) of the local time-activity curves Q(t). This index defined as: 1"120
1"120
COG12o= J o ta(t)dt/J o a(t)dt
and so:
COG12o(R) = (2)
f
(tT(t)+tE(t))dt
~
.
Jo (T(t)+E(t))dt
Intravenous radionuclide angiography• A. BARTOLINIet al.
93
Table 1. Mean value of regional COG120 and of side-to-side difference in controls, in patients with cerebral ischemia, intracranial tumors and arteriovenous angiomas.
Patients
Controls 15
Ischemic Patients 23
Tumors 6
Av angioma 2
SPECT ratio 1.03 -+ 0.03 1.32 - 0.25 1.2 _ 0.08 1.18 ± 0.03 COG mean value (sec) 72.75 --- 5.33 72.9 --- 4.9 73.01 _+ 4.6 72.1 --- 3.8 COG Side-to-side difference (sec) 0.5 ± 0.13 0.6 --- 0.11 1.1 ___0.08 2.2 ± 0.06 COG side-to-side difference exceeding confidence limits (N of patients)
0
3
2
This value is unaffected by blood flow depending solely upon the integration time (120 sec), and upon the relation between T(t) and E(t). Its values range from a minimum, in the absence of extravasation (E(t)= 0), related to intravascular circulation time, and increases progressively with the weight of E(t) with respect to T(t) to a maximum for total uptake (T(t)= 0) corresponding to 1/2 (120 sec minus the starting time of brain activity). By computing the difference between COG12o values in symmetrical regions, the effects due to I(t) are erased, and we obtain the difference between the respective centers of gravities of R(O. This difference may offer a convenient way to compare the similarity of the backfluxes in regions showing a different uptake by SPECT. Symmetry of COG12o guarantees similarity of backflux and of those factors affecting 99mTc HM-PAO uptake which are not related to blood flow; an increased or decreased COG120 value will reflect an increased or decreased tissue retention, both independent from blood flow.
Data analysis Side-to-side differences of COG120 values were assessed, using 12 symmetrical ROIs on each hemisphere, both in control subjects and in patients. Particular attention was paid to the COG values corresponding to SPECT decreased uptake and to the contralateral normal region. RESULTS The table summarizes the results. In normal subjects mean regional COG12 o was 72.75
Fig. 1. COG120 symmetrical image in a ischemic patient with a conspicuous SPECT decreased uptake (R: fight side; L: left side; P: posterior side).
- 5.33 sec; mean side-to-side difference (higer-lower value) was 0.5 ___ 0.13 sec; confidence limits (mean ___ 2 SD) for side-to-side differences were 0.76 seconds. There was no significant difference between patients with cerebral ischemia and normal subjects with regard to either mean value of COGle o or mean side-to-side difference between regions corresponding to changed S P E C T uptake and contralateral regions; in no patient did COG12o side-to-side difference exceed normal confidence limits. A typical case is illustrated in Fig. 1. Patients with nonischemic lesions (6 tumors and 2 av
ComputerizedMedicalImaging and Graphics
94
A
March-April/1990, Volume 14, Number2
B
E = (1 - e -
PS/F)
(8)
(P=permeability, S=vascular surface) and
Uptake=FE yro Ca(t)dt.
Fig. 2. Changes of local COG12o in nonischemic lesions with changed SPECT uptake. A: Meningioma of the olfactory groove with decreased COG12o value and SPECT increased uptake (arrow); B: Left arteriovenous angioma with decreased COGazo value and paradoxical SPECT decreased uptake (arrows) (R: right side; L: left side; P: posterior side).
angiomas) had the same COG12o mean value as normal subjects; however, side-to-side differences exceeding normal confidence limits were found in all av angiomas and in 3 meningiomas (2 with decreased, 1 with increased SPECT uptake) (Fig. 2). In all these cases COG12o was decreased on the side corresponding to the lesion. DISCUSSION For a correct rCBF qualitative estimation by 99mTc HM-PAO uptake, all parameters related to bbb permeability and tissue 99mTcHM-PAO retention in all regions of the same patients must be similar. We have assumed that possible local changes in any parameter other than blood flow could determine a change in the relation transmitted/extravasated activity and that this could be detected by the corresponding change in COG~2o. The relation between transmitted and extravasated activity is governed by the value of brain 99mTc HMPAO extraction and by the extent of the initial backflux. A tracer is taken up initially by the tissue according to the product of blood flow (F), arterial concentration (Ca) and extraction (E) defined as (9):
(9)
Linearity between blood flow and uptake is ensured only for values of the ratio PS/F which are sufficiently high for us to consider E to be independent from blood flow both in normal and pathological conditions. Our results can be summarized as follows: (a) In normal subjects side-to-side differences in COGla o show a sufficiently small range, corresponding to about 1% of the absolute value of COG12o, (b) none of the 23 ischemic patients with decreased SPECT uptake showed a corresponding side-to-side difference in COG12o. This indicates that the relation between the transmitted and extravasated fraction was similar in the region of decreased uptake and the controlateral normal region. The suggestion here is that, although the actual value of the extraction coefficient E is unknown, the value PS (permeability surface product) was sufficiently high to warrant a similarity of the factor of extraction in normal and ischemic tissue, in spite of conspicuous differences in blood flow. Since this pattern was found in all 23 patients with ischemia one may conclude that the influence of factors other than blood flow in determining the occurrence of changed SPECT uptake may be taken as marginal and possibly restricted to individual cases, (c) all changes in COG12o were found in correspondence to nonischemic lesions (2 arteriovenous angiomas, 3 meningiomas), and were all characterized by a decreased value entailing a decreased retention and an underestimation of blood flow by SPECT 99mTc HM-PAO uptake. In av angiomas, the decreased COG12o indicates a prevalence of T(t) over E(t) and this decreased retention determined a paradoxical decreased SPECT uptake. In meningiomas the decreased COG12o was associated to both increased and decreased SPECT uptake, again indicating an underestimation of flow due to decreased retention. In conclusion, events other than flow which occur in the first 120 sec after injection in cerebral ischemic patients do not substantially affect 99mTc HM-PAO retention so that in normal subjects and ischemic patients the linearity of flow uptake may be safely assumed. The decreased retention found in arteriovenous angiomas and meningiomas suggests the possible occurrence of some of the following conditions in which changed parameters other than blood flow may affect 99mTc
Intravenous radionuclide angiography • A. BARTOL~Xqet al.
HM-PAO uptake: (a) hyperhaemic lesions which, like angiomas, may entail a decreased extraction value; (b) lesions consisting of tissue other than brain, such as meningiomas; this suggests that the factors involved in 99mTC HM-PAO retention are linked to cerebral tissue. Radionuclide angiography with computer generated COGt2 o images can be a fairly simple but useful adjunct to standard SPECT examination. It may provide information which allows us to state whether or not a changed uptake detected SPECT is related to the corresponding change in blood flow. SUMMARY Cerebral radionuclide angiography (RA), following intravenous administration of 99mTC HM-PAO, was performed for 120 sec in a group of patients with cerebral ischemia and brain tumors prior to SPECT performance. Sequential images were elaborated in order to obtain a planar representation of the first normalized mathematical moment or center of gravity (COG120) up to 120 sec of the local time activity curves. Since this index is independent of cerebral blood flow and affected solely by the relation between transmitted and extracted activity, it was deemed suitable for detecting differences in 99mTc HM-PAO retention independent of blood flow in regions analyzed for differences in blood flow by SPECT. Side-to-side differences in COG12 o were compared to side-to-side ratios of SPECT activity. All 23 patients with cerebral ischemia showed conspicuous side-to-side differences at SPECT without a c o r r r e s p o n d i n g COG120 difference. Five patients with nonischemic lesions (3 with a meningioma and 2 with an arteriovenous angioma) showed a changed SPECT uptake associated to a significantly decreased COG12o value on the injured side. It was concluded that in ischemia, when a decreased flow is expected, 99mTc HM-PAO uptake and rCBF are linearly related; in hyperhaemia or in lesions of extracerebral origin a decreased tissue retention independent of blood flow may occur. Radionuclide angiography with 99mTc HM-PAO, easily performed prior to SPECT, may be useful in allowing us to check whether a detected asymmetry at SPECT may be taken as corresponding to a changed blood flow. REFERENCES 1. V01kert, W.A.; Hoffman, T.J.; Seger, R.M.; Troutuer, D.E.; Holmes, R.A. Tc99mpropyleneamine oxime (Tc99mpnAO) a poten-
95
tial brain radiopharmaceutical. Eur. J. Nucl. Med. 9:511; 1984. 2. Holmes, R.A.; Chaplin, S.B.; Royston, K.J.; Hoffman, T.; Volkert, W.A. Cerebral uptake and retention of 99mTChexamethylpropyleneamineoxime (99mTc HM-PAO). Nucl. Med. Commun. 6:443; 1985. 3. Novotnik, D.P.; Canning, R.L.; Cumming, S.A. Development of a Tc99mlabelled radiopharmaceutical for cerebral blood flow imaging. Nucl. Med. Commun. 6:499; 1985. 4. Sharp, P.F.; Smith, H.G.; Gemmel, H.G.; Lyall, D.; Evans, N.; Gvozdanovic, D.; Davidson, J.; Tyrrel, D.; Pickett, R.; Neirinckx, R. Technetium99m HM-PAO stereoisomers as potential agents for imaging regional cerebral blood flow: human volunteer studies. J. Nucl. Med. 27:171; 1986. 5. Ell, P.J.; Hocknell, J.M.L.; Jarrit, P.H.; Ullum, I.; Lui, D.; Campos-Costa, D.; Novotnik, D.; Pickett, R.; Canning, L.; Neirinckx, R. A 99mTClabelled radiotracer for the investigation of cerebrovascular disease. Nucl. Med. Commun. 6:437; 1985. 6. Leonard, J.P.; Novotnik, D.P.; Neirinckx, R.D. Technetium 99mTc HM-PAO: a new radiopharmaceutical for imaging regional brain perfusion using SPECT. A comparison with Iodine 123 HIPDM. J. Nucl. Med. 27:1818; 1986. 7. Bartolini, A.; Gasparetto, B.; Bacigalupo, F.; Ruffinengo, U.; Amore, R.; Loeb, C. SPECT with 99mTCHM-PAO in the clinical assessment of cerebral ischemia: a preliminary evaluation. Eur. Neurol. 28:232; 1988. 8. Bartolini, A. Regional Arm Brain mean transit time in the diagnostic of patients with cerebral vascular disease. Stroke 12:241; 1981. 9. Lassen, N.A.; Perl, W. Tracer kinetics methods. In: Medical Physiology. New York: Raven Press; 1979:113.
About the Author--ALFREOOBARTOLINI,MD received his degree from the University of Genoa (Italy) in 1965. In 1965 he joined file Department of Clinical Neurology of the University of Genoa. He is presently Associate Professor of Neuroradiology at the University of Genoa. His research activities center around radioisotope assessment of cerebral vascular disease. About the Author--BRuNO GASPARETTOreceived the Italian degree in
Electronic Engineering from the University of Genoa (Italy) in 1972. In 1970 he joined the Center of Cerebral Neurophysiology of the CNR (National Research Council of Italy), where he is now first researcher. His research interests include modeling physiological systems, signal processing and image processing. About the Author--UBERTO Rur~n~.N60 received his MD degree from
the University of Genoa (Italy) in 1984. He is a specialist in Clinical Neurology. He presently serves on the Neuroradiological Service at the Galliera Hospital of Genoa. About the Author--RAtAbLE AMOREis a graduate in Chemistry and
works as an assistant at the CNR (National Research Council of Italy), where he started his activity in 1963. He joined the Center for Cerebral Neurophysiology of the CNR in 1970. About the Author--ALBERTO ~ V E R A ,
MD received his degree from the University of Genoa (Italy) in 1972. He is a specialist in Neurology, Neurophysiopathology and Anesthesiology. His major field of interest is in the Intensive Care for Neurological Disease. About the Author--CLAUDIO ALBANO,MD received his degree from the University of Genoa (Italy) in 1973. He is a specialist in Neurology, Physiotherapy and Clinical Pharmacology. He is presently Associate Professor of Neurological Therapy at the University of Genoa. His current research interests are centered around Neuropharmacology.
0895-6111/90 $3.00 + .00 Copyright © 1990 Pergamon Press plc
Printed in the USA. All rights reserved.
INTRAVENOUS 99mTCHM-PAO RADIONUCLIDE ANGIOGRAPHY AS AN ADJUNCT TO SPECT ASSESSMENT OF CEREBRAL DISEASE A. Bartolini, B. Gasparetto, U. Ruffinengo, R. Amore, A. Prima'cera and C. Albano Institute of Clinical Neurology, University of Genoa CNR-Center of Cerebral Neurophysiology, Genoa, Italy
(Received24 March 1989;Revised 10 August 1989) Abstract--An application of intravenous radionuclide angiography as an adjunct to SPECT with 99roTe HM-PAO is presented. From the angiographic sequence a two-dimensional parametric image is generated computing, pixel by pixel, the center of gravity (COG) of the local time curves. This index is independent of blood flow and reflects the relation between transmitted and extravasated activity. COG value, compared in two symmetrical regions analyzed for differences in flow by SPECT, may help to state whether a detected SPECT changed uptake is linearly related or not to the corresponding change of blood flow. All 23 ischemic patients taken into consideration had a similar backflux in symmetrical regions with conspicuous changed uptake. In 3 meningiomas and 2 arteriovenous angiomas it was possible to detect the occurrence of a changed backflux in the lesion suggesting a nonlinearity between blood flow and 99mTCHM-PAO uptake.
Key Words: SPECT, 99mTcHM-PAO, Radionuclide angiography, Cerebral ischemia, Cerebral blood flow metabolic activity. Therefore the statement that the spatial distribution of 99mTCHM-PAO uptake is linearly related to that of blood flow is true only on condition that possible local changes in the permeability and tissue parameters in brain lesions do not determine substantial additional changes in 99mTcHM-PAO retention. Steady state 99mTc HM-PAO uptake, Qr of a monitored cerebral region, may be described as the product of flow, F, and integrated arteriovenous differences:
INTRODUCTION
99m-Technetium hexamethylpropyleneamineoxime (99rnTc HM-PAO) has been reported to cross the intact blood-brain barrier and to distribute in the brain in proportion to regional blood flow. Moreover it is retained sufficiently long to allow single photon emission computed tomography (SPECT) with widely available rotating gamma camera systems (1-6). At present, the lack of conclusive data concerning the fate of 99mTcHM-PAO in the brain does not allow rCBF to be quantified. A simple qualitative analysis based on the expected symmetry of 99rnTc HM-PAO distribution is, however, possible. Although the reliability of 99mTcHM-PAO uptake as a qualitative steady state indicator of cerebral blood flow has been supported by the high incidence of decreased 99mTcHM- PAO uptake in the side of brain ischemia (7), the possibility that factors other than blood flow may influence the actual 99mTc HM-PAO uptake cannot be ruled out. Generally speaking, the steady state uptake of 99mTc HM-PAO depends, in addition to blood flow, upon blood-brain barrier (bbb) permeability and mean residence time of the tracer within tissue. This last factor is governed by tissue affinity for 99mTc H/VI-PAO, and possibly reflects cellular and chemical composition and
Qr=F yf(C2(t)-Cv(t))dt
(l)
(Ca(t) and Cv(t)= arterial and venous concentrations). Assuming that integrated arterial activity is similar in all regions of the same patient, any change in uptake independent from blood flow will be associated with a change in the venous concentration curve. Radionuclide angiography may be easily performed following bolus injection of 99myc HM-PAO prior to SPECT examination. Analysis of local curves may provide, although on a simple planar basis, a way to compare in any patient the time course of the activity in regions analyzed for differences of flow by SPECT. The following study illustrates an application of radionuclide angiography, based on computer generation of center of gravity (COG) images, as a useful adjunct to the SPECT study. Our aim was to discover whether this simple, inexpensive technique was suitable for detecting regional changes in backflux and to what extent a
Address correspondenceto: ProfessorAlfredoBartolini, Centro di Neurofisiologia Cerebrale, Clinica Neurologica dell'Universita', Via De Toni, 5 16132Genova, Italy. 91
92
Computerized Medical Imagingand Graphics
changed backflux could explain a changed SPECT uptake.
PATIENTS AND METHODS
Patients Fifteen control subjects and 31 patients with a conspicuously changed 99mTC HM-PAO uptake were selected. The 15 control subjects underwent 99mTCHMPAO SPECT as a screening procedure for minor symptoms; this proved negative. In all subjects clinical observation and CT examination allowed us to rule out major cerebral disease. Of the 31 patients with changed 99mWC HM-PAO SPECT uptake, 29 had a decreased uptake and 2 an increased uptake; 23 suffered from cerebral ischemia clinically presenting with a completed stroke dating from 5 to 30 days prior to SPECT examination, 2 patients had an arterio-venous (av) angioma, and 6 had an intracranial tumor (4 meningiomas, 1 glioma, 1 metastasis). Diagnoses were based on the usual workup, including CT in all patients, and cerebral angiography when diagnostically necessary. All patients with changed SPECT uptake had an asymmetrical distribution of the tracer on visual inspection of sequential radionuclide angiography images.
March-April/1990, Volume 14, Number2
was calculated about the injection time and up to 120 secs. The choice of this time was based on the observation, reported by Leonard et al. (6), that the brain activity following intravenous injection slowly changes only in this interval. An image representing the distribution of COG12o was generated by summing up the original 120 frames, having multiplied each by the corresponding time delay from injection. The picture obtained was then divided by the sum of the original 120 frames. Externally-monitored brain time-activity curve Q(t) results from the product of blood flow (F) and the convolution of I(t), input function, with R(t) residue function or local unit impulse response (8,9):
(3)
Q(t) =Ff2I(u)(t-u)du.
I(t) represents the time-concentration curve crossing the local inflow and R(t) corresponds to the function which would have been recorded on the tissue for unit impulsive administration at the cerebral inflow. Using the Laplace transform, the COG120 of the local cerebral intravenous curve can be broken down as (8,9): 1"120
1"120
COG120=J0 tQ(t)dt/J o Q(t)dt=
Methods Radionuclide angiography was performed by monitoring the passage through the brain of a bolus of about 15 mCi of 99mTCHM-PAO injected into the right brachial vein. Following injection, a sequence of 120 frames at 1 sec intervals was started, with the patients positioned in the supine position, facing the detector. SPECT was performed 10 minutes later, using a rotating gamma camera (GE 400A) connected on line to a computer (GE STAR). The examination was carded out by taking 64 views during 360 degree rotation about the long axis of the patient and accumulating about 5 million counts for each examination. Tomographic reconstruction was performed by filtered back-projection (Ramp Hanning filter, cut-off frequency = 0.5 cycles/ pixel) and attenuation correction (K= 0.07).
120
1"120
1"120
1"120
tli(t)dt/J o I(t)dt+ J o tR(t)/J o R(t)dt
(4)
COG12o= COG12o(I)+ COG12o(R).
(5)
and
Assuming I(t) to be similar in all regions of the same patient, the regional COG12o values may differ among themselves only in the component COG12o (R). But R(t) may be described as resulting from the sum of intravascular transmitted activity T(t) and extravasated activity E(t) (9):
R(t) = T(t) + E(t)
(6)
2°
(7)
Theoretical considerations Sequential radionuclide angiography flames were elaborated with a view to obtaining a two-dimensional representation of the center of gravity (COG12o) of the local time-activity curves Q(t). This index defined as: 1"120
1"120
COG12o= J o ta(t)dt/J o a(t)dt
and so:
COG12o(R) = (2)
f
(tT(t)+tE(t))dt
~
.
Jo (T(t)+E(t))dt
Intravenous radionuclide angiography• A. BARTOLINIet al.
93
Table 1. Mean value of regional COG120 and of side-to-side difference in controls, in patients with cerebral ischemia, intracranial tumors and arteriovenous angiomas.
Patients
Controls 15
Ischemic Patients 23
Tumors 6
Av angioma 2
SPECT ratio 1.03 -+ 0.03 1.32 - 0.25 1.2 _ 0.08 1.18 ± 0.03 COG mean value (sec) 72.75 --- 5.33 72.9 --- 4.9 73.01 _+ 4.6 72.1 --- 3.8 COG Side-to-side difference (sec) 0.5 ± 0.13 0.6 --- 0.11 1.1 ___0.08 2.2 ± 0.06 COG side-to-side difference exceeding confidence limits (N of patients)
0
3
2
This value is unaffected by blood flow depending solely upon the integration time (120 sec), and upon the relation between T(t) and E(t). Its values range from a minimum, in the absence of extravasation (E(t)= 0), related to intravascular circulation time, and increases progressively with the weight of E(t) with respect to T(t) to a maximum for total uptake (T(t)= 0) corresponding to 1/2 (120 sec minus the starting time of brain activity). By computing the difference between COG12o values in symmetrical regions, the effects due to I(t) are erased, and we obtain the difference between the respective centers of gravities of R(O. This difference may offer a convenient way to compare the similarity of the backfluxes in regions showing a different uptake by SPECT. Symmetry of COG12o guarantees similarity of backflux and of those factors affecting 99mTc HM-PAO uptake which are not related to blood flow; an increased or decreased COG120 value will reflect an increased or decreased tissue retention, both independent from blood flow.
Data analysis Side-to-side differences of COG120 values were assessed, using 12 symmetrical ROIs on each hemisphere, both in control subjects and in patients. Particular attention was paid to the COG values corresponding to SPECT decreased uptake and to the contralateral normal region. RESULTS The table summarizes the results. In normal subjects mean regional COG12 o was 72.75
Fig. 1. COG120 symmetrical image in a ischemic patient with a conspicuous SPECT decreased uptake (R: fight side; L: left side; P: posterior side).
- 5.33 sec; mean side-to-side difference (higer-lower value) was 0.5 ___ 0.13 sec; confidence limits (mean ___ 2 SD) for side-to-side differences were 0.76 seconds. There was no significant difference between patients with cerebral ischemia and normal subjects with regard to either mean value of COGle o or mean side-to-side difference between regions corresponding to changed S P E C T uptake and contralateral regions; in no patient did COG12o side-to-side difference exceed normal confidence limits. A typical case is illustrated in Fig. 1. Patients with nonischemic lesions (6 tumors and 2 av
ComputerizedMedicalImaging and Graphics
94
A
March-April/1990, Volume 14, Number2
B
E = (1 - e -
PS/F)
(8)
(P=permeability, S=vascular surface) and
Uptake=FE yro Ca(t)dt.
Fig. 2. Changes of local COG12o in nonischemic lesions with changed SPECT uptake. A: Meningioma of the olfactory groove with decreased COG12o value and SPECT increased uptake (arrow); B: Left arteriovenous angioma with decreased COGazo value and paradoxical SPECT decreased uptake (arrows) (R: right side; L: left side; P: posterior side).
angiomas) had the same COG12o mean value as normal subjects; however, side-to-side differences exceeding normal confidence limits were found in all av angiomas and in 3 meningiomas (2 with decreased, 1 with increased SPECT uptake) (Fig. 2). In all these cases COG12o was decreased on the side corresponding to the lesion. DISCUSSION For a correct rCBF qualitative estimation by 99mTc HM-PAO uptake, all parameters related to bbb permeability and tissue 99mTcHM-PAO retention in all regions of the same patients must be similar. We have assumed that possible local changes in any parameter other than blood flow could determine a change in the relation transmitted/extravasated activity and that this could be detected by the corresponding change in COG~2o. The relation between transmitted and extravasated activity is governed by the value of brain 99mTc HMPAO extraction and by the extent of the initial backflux. A tracer is taken up initially by the tissue according to the product of blood flow (F), arterial concentration (Ca) and extraction (E) defined as (9):
(9)
Linearity between blood flow and uptake is ensured only for values of the ratio PS/F which are sufficiently high for us to consider E to be independent from blood flow both in normal and pathological conditions. Our results can be summarized as follows: (a) In normal subjects side-to-side differences in COGla o show a sufficiently small range, corresponding to about 1% of the absolute value of COG12o, (b) none of the 23 ischemic patients with decreased SPECT uptake showed a corresponding side-to-side difference in COG12o. This indicates that the relation between the transmitted and extravasated fraction was similar in the region of decreased uptake and the controlateral normal region. The suggestion here is that, although the actual value of the extraction coefficient E is unknown, the value PS (permeability surface product) was sufficiently high to warrant a similarity of the factor of extraction in normal and ischemic tissue, in spite of conspicuous differences in blood flow. Since this pattern was found in all 23 patients with ischemia one may conclude that the influence of factors other than blood flow in determining the occurrence of changed SPECT uptake may be taken as marginal and possibly restricted to individual cases, (c) all changes in COG12o were found in correspondence to nonischemic lesions (2 arteriovenous angiomas, 3 meningiomas), and were all characterized by a decreased value entailing a decreased retention and an underestimation of blood flow by SPECT 99mTc HM-PAO uptake. In av angiomas, the decreased COG12o indicates a prevalence of T(t) over E(t) and this decreased retention determined a paradoxical decreased SPECT uptake. In meningiomas the decreased COG12o was associated to both increased and decreased SPECT uptake, again indicating an underestimation of flow due to decreased retention. In conclusion, events other than flow which occur in the first 120 sec after injection in cerebral ischemic patients do not substantially affect 99mTc HM-PAO retention so that in normal subjects and ischemic patients the linearity of flow uptake may be safely assumed. The decreased retention found in arteriovenous angiomas and meningiomas suggests the possible occurrence of some of the following conditions in which changed parameters other than blood flow may affect 99mTc
Intravenous radionuclide angiography • A. BARTOL~Xqet al.
HM-PAO uptake: (a) hyperhaemic lesions which, like angiomas, may entail a decreased extraction value; (b) lesions consisting of tissue other than brain, such as meningiomas; this suggests that the factors involved in 99mTC HM-PAO retention are linked to cerebral tissue. Radionuclide angiography with computer generated COGt2 o images can be a fairly simple but useful adjunct to standard SPECT examination. It may provide information which allows us to state whether or not a changed uptake detected SPECT is related to the corresponding change in blood flow. SUMMARY Cerebral radionuclide angiography (RA), following intravenous administration of 99mTC HM-PAO, was performed for 120 sec in a group of patients with cerebral ischemia and brain tumors prior to SPECT performance. Sequential images were elaborated in order to obtain a planar representation of the first normalized mathematical moment or center of gravity (COG120) up to 120 sec of the local time activity curves. Since this index is independent of cerebral blood flow and affected solely by the relation between transmitted and extracted activity, it was deemed suitable for detecting differences in 99mTc HM-PAO retention independent of blood flow in regions analyzed for differences in blood flow by SPECT. Side-to-side differences in COG12 o were compared to side-to-side ratios of SPECT activity. All 23 patients with cerebral ischemia showed conspicuous side-to-side differences at SPECT without a c o r r r e s p o n d i n g COG120 difference. Five patients with nonischemic lesions (3 with a meningioma and 2 with an arteriovenous angioma) showed a changed SPECT uptake associated to a significantly decreased COG12o value on the injured side. It was concluded that in ischemia, when a decreased flow is expected, 99mTc HM-PAO uptake and rCBF are linearly related; in hyperhaemia or in lesions of extracerebral origin a decreased tissue retention independent of blood flow may occur. Radionuclide angiography with 99mTc HM-PAO, easily performed prior to SPECT, may be useful in allowing us to check whether a detected asymmetry at SPECT may be taken as corresponding to a changed blood flow. REFERENCES 1. V01kert, W.A.; Hoffman, T.J.; Seger, R.M.; Troutuer, D.E.; Holmes, R.A. Tc99mpropyleneamine oxime (Tc99mpnAO) a poten-
95
tial brain radiopharmaceutical. Eur. J. Nucl. Med. 9:511; 1984. 2. Holmes, R.A.; Chaplin, S.B.; Royston, K.J.; Hoffman, T.; Volkert, W.A. Cerebral uptake and retention of 99mTChexamethylpropyleneamineoxime (99mTc HM-PAO). Nucl. Med. Commun. 6:443; 1985. 3. Novotnik, D.P.; Canning, R.L.; Cumming, S.A. Development of a Tc99mlabelled radiopharmaceutical for cerebral blood flow imaging. Nucl. Med. Commun. 6:499; 1985. 4. Sharp, P.F.; Smith, H.G.; Gemmel, H.G.; Lyall, D.; Evans, N.; Gvozdanovic, D.; Davidson, J.; Tyrrel, D.; Pickett, R.; Neirinckx, R. Technetium99m HM-PAO stereoisomers as potential agents for imaging regional cerebral blood flow: human volunteer studies. J. Nucl. Med. 27:171; 1986. 5. Ell, P.J.; Hocknell, J.M.L.; Jarrit, P.H.; Ullum, I.; Lui, D.; Campos-Costa, D.; Novotnik, D.; Pickett, R.; Canning, L.; Neirinckx, R. A 99mTClabelled radiotracer for the investigation of cerebrovascular disease. Nucl. Med. Commun. 6:437; 1985. 6. Leonard, J.P.; Novotnik, D.P.; Neirinckx, R.D. Technetium 99mTc HM-PAO: a new radiopharmaceutical for imaging regional brain perfusion using SPECT. A comparison with Iodine 123 HIPDM. J. Nucl. Med. 27:1818; 1986. 7. Bartolini, A.; Gasparetto, B.; Bacigalupo, F.; Ruffinengo, U.; Amore, R.; Loeb, C. SPECT with 99mTCHM-PAO in the clinical assessment of cerebral ischemia: a preliminary evaluation. Eur. Neurol. 28:232; 1988. 8. Bartolini, A. Regional Arm Brain mean transit time in the diagnostic of patients with cerebral vascular disease. Stroke 12:241; 1981. 9. Lassen, N.A.; Perl, W. Tracer kinetics methods. In: Medical Physiology. New York: Raven Press; 1979:113.
About the Author--ALFREOOBARTOLINI,MD received his degree from the University of Genoa (Italy) in 1965. In 1965 he joined file Department of Clinical Neurology of the University of Genoa. He is presently Associate Professor of Neuroradiology at the University of Genoa. His research activities center around radioisotope assessment of cerebral vascular disease. About the Author--BRuNO GASPARETTOreceived the Italian degree in
Electronic Engineering from the University of Genoa (Italy) in 1972. In 1970 he joined the Center of Cerebral Neurophysiology of the CNR (National Research Council of Italy), where he is now first researcher. His research interests include modeling physiological systems, signal processing and image processing. About the Author--UBERTO Rur~n~.N60 received his MD degree from
the University of Genoa (Italy) in 1984. He is a specialist in Clinical Neurology. He presently serves on the Neuroradiological Service at the Galliera Hospital of Genoa. About the Author--RAtAbLE AMOREis a graduate in Chemistry and
works as an assistant at the CNR (National Research Council of Italy), where he started his activity in 1963. He joined the Center for Cerebral Neurophysiology of the CNR in 1970. About the Author--ALBERTO ~ V E R A ,
MD received his degree from the University of Genoa (Italy) in 1972. He is a specialist in Neurology, Neurophysiopathology and Anesthesiology. His major field of interest is in the Intensive Care for Neurological Disease. About the Author--CLAUDIO ALBANO,MD received his degree from the University of Genoa (Italy) in 1973. He is a specialist in Neurology, Physiotherapy and Clinical Pharmacology. He is presently Associate Professor of Neurological Therapy at the University of Genoa. His current research interests are centered around Neuropharmacology.