Clinical Investigative Studies

Cerebral Perfusion Index
A New Marker for Clinical Outcome in
Acute Stroke

Andrei V. Alexandrov, MD

Christopher F. Bladin, MD

Stroke Research Unit
Sunnybrook Health Science
Centre
University of Toronto
Toronto, Canada

Stroke Research Unit
Sunnybrook Health Science
Centre

Lisa E. Ehrlich, MD
Department of Nuclear
Medicine
Sunnybrook Health Science
Centre

ABSTRACT

John W. Norris, MD, FRCP
Stroke Research Unit
Sunnybrook Health Science
Centre

Single-photon emission computed tomography (SPECT) and

cardioembolic stroke was found. The three grades of the initial
CPI (1-5, 6-12, 15-20) were found to predict different de-

transcranial Doppler (TCD) ultrasound are of good prognostic

grees of short-term outcome (good, medium, and poor). The

value in acute stroke, and combined they may be an accurate

TCD and SPECT combination (CPI) may prove to be a safe, fast,

way to determine a target group of patients with maximum

and reliable noninvasive substitute for angiography.

therapeutic response. Seventy consecutive patients were studied (42 with middle cerebral artery strokes, 18 with transient
ischemic attacks [TIAs]; 10 were excluded due to failure of insonation). Two SPECT studies were performed at 2.1 ± 1.2

Alexandrov AV. Ehrlich LE, Bladin CF, Norris JW. Cerebral
perfusion index: a new marker for clinical outcome in
acute stroke. J Neuroimag 1993;3:209-215

and 13.8 ± 3.1 days after onset. Serial TCD studies were done
at 10 hours and at the time of the SPECT studies. Neurological
deficit was scored on admission and 2 weeks later (using the
Canadian Neurological Scale). Cerebral perfusion index (CPI)
was derived by multiplying the values for TCD and SPECT patterns. Positive correlation was obtained in all 16 patients in
whom cerebral angiography was performed within the first 3
days after onset. The occlusive TCD pattern and absence of
perfusion on SPECT were common in the stroke group (19/42
patients) and were never seen in those with TIAs. A normal
TCD pattern and normal perfusion on SPECT were more common in the patients with TIAs (9/18 vs 8/42, p = 0.02; 5/10
vs 1/40, p = 0.0003). The occlusive TCD and SPECT patterns
were associated with the highest mean infarction volume (147

± 87 vs 19 ± 21, p < 0.0001) and all nonocclusive TCD and
SPECT patterns were associated with the better short-term
outcome (43.2 ± 33.9 vs 92.4 ± 20.2, p < 0.0001). A tendency for hyperfixation of the tracer to be more common in

Received Nov 13, 1992, and in revised form Feb 2, May 7,
and May 21, 1993. Accepted for publication May 22, 1993.
Address correspondence to Dr Alexandrov, Stroke Research
Unit, 2075 Bayview Ave, Toronto, Canada M4N 3M5.

At present, there is no definitive treatment for acute ischemic stroke; cytoprotective and thrombolytic therapies
are the most promising [1]. Cerebral angiography and
computed tomography (CT) are used in the immediate
phase of stroke to determine patients suitable for intraarterial or intravenous thrombolysis [2-4]. Arteriography
is the "gold standard" but is invasive, time-consuming,
and impractical in many cases. CT is not sensitive
enough to show cerebral ischemia within the first hours
[5, 6].
Single-photon em1sswn computed tomography
(SPECT) and transcranial Doppler (TCD) ultrasound allow noninvasive assessment of cerebral perfusion and are
informative and of good predictive value in acute stroke
[7-10]. Since acute thrombus formation and arterial occlusion are present in more than 70% of stroke patients
within the first 6 hours [11], a combination ofTCD and
SPECT studies in the immediate phase may be an accurate and informative way of determining a target group
of patients for maximum thrombolytic efficacy. Therefore, acute study to correlate serial TCD and SPECT measurements in acute and subacute phases of cerebral
ischemia was carried out.

Copyright © 1993 by the American Society of Neuroimaging

209

STENOTIC

NORMAL

Subjects and Methods

Consecutive patients with symptoms of middle cerebral
artery (MCA) stroke admitted to the emergency department within 24 hours after onset were studied. CT was
performed twice: on admission and at 3 to 4 days after
onset. The volume of the ischemic lesion was measured
by a special software program (Sigma Scan, Jandel
Scientific, Corte Madera, CA). Two SPECT studies
were performed in the subacute period (days 1-3 and 14)
using technetium 99m-hexamethylpropyleneamineoxime (HMPAO, Amersham, UK). Serial TCD measurements were performed on admission and at the time of
the SPECT studies (TC-2-64 EME). For SPECT, the following patterns of fixation ofHMPAO were used: normal,
high, mixed, low, and severe perfusion deficit (SPD) (i.e.,
absence of perfusion); for TCD, the following were used:
normal, collateral, stenotic, and occlusive (Fig 1). An index representing cerebral perfusion was constructed by
combining TCD and SPECT results. TCD reflects patency of arteries in the circle of Willis, and the authors'
own comparison with angiographic controls and that of
others [12, 13] indicate it is fairly accurate. SPECT reflects perfusion of brain parenchyma and previous reports indicate that this is also accurate [8, 10]. By combining the two techniques, the accuracy of diagnosis of
perfusion deficits and the reason for their presence may
be increased. The term "cerebral perfusion index" (CPI)
is used (Table 1). Neurological deficit was scored by
means of the Canadian Neurological Scale (CNS) [14]
on admission and 2 weeks later. The risk of cardiogenic
embolism was estimated using the Toronto Embolic
Scale (TES) [ 15], which combines clinical and laboratory
findings.
TCD reports were analyzed independently of other
clinical and laboratory data, SPECT results were reported
by a neuroradiologist (L. E. E.) who was unfamiliar with
the clinical findings, and scoring of neurological deficit
was performed by a neurologist (C. F. B.) who was unaware of the TCD and SPECT results.
Carotid duplex sonography was performed in all patients. No patient was treated with thrombolytic agents.
Results

Seventy consecutive patients were studied: 10 were ex-

COLLATERAL

OCCLUSIVE

Fig 1. TCD patterns of flow direction and velocities of the
circle of Willis in acute stroke.

eluded from the study because of technical failure of ultrasound insonation or because they were uncooperative;
42 had completed strokes (mean age, 73.6 ± 11.9 yr; 25
men, 17 women) and 18 had hemispheric TIAs (73.8 ±
12.1 yr; 9 men, 9 women). In the stroke group, 5 patients
died within the first 2 weeks. The mean time of the initial
TCDwas 10.3 ± 4.1 hours(strokes)and 10.1 ± 6.2hours
(TIAs) correspondingly. Mean time of the first SPECT
and interim TCD studies was 2.1 ± 1.2 days. The second
SPECT and "final" TCD .studies were done at 13.8 ±
3.1 days after onset. In the TIA group, only the first TCD/

Table 1. Cerebral Perfusion Index (CPI): SPECT Pattern X TCD Pattern
TCD Values

.210

Patterns

4
(Normal)

3
(Collateral)

2
(Stenotic)

(Occlusive)

SPECT values
5 (normal)
4 (high)
3 (mixed)
2 (low)
1 (absence of perfusion)

20
16
12
8
4

15
12
9
6
3

10
8
6
4
2

5
4
3
2
1

Journal of Neuroimaging Vol3 No 4 October 1993

1

Table 2. TCD and SPECT Perfusion Patterns in
Stroke and TIA Groups
Initial
(TIA)

Patterns

--

90 -

Final

Interim

-

CNS

70 -

TCD
Normal
Collateral
Stenotic
Occlusive
Total

8 (9)
4 (3)
II (6)
19 (0)
42 (18)

II
6
17
6
40

SPECT
Normal
High
Low
Absence of perfusion
Total

15
8
13
2

-,

40 -

-

,-- -

-r-

(0) n=19
i nitial CN!

f--

38.9+/-26
final CNS:
43,2+/-33

-'-

--

(NO) n=23
initial CN

- f--

-

r-

-

77.5+1-2

38

10 -

0

I (5)
9 (3)
II (2)
19 (0)
40 (10)

--

ll
18
2
31

0

[\()

',

p<0.0001

final CNS:
92.4+/-2

a'-

[\()

p<0.0001

Fig 2. Canadian Neurological Scale (CNS) scores in occlusi
(0) and nonocclusion (NO) groups at first 10 hours and at 2 we
after onset.

SPECT study was available: The neurological deficit resolved in the majority of TIA patients at the time of the
initial TCD study and they were not admitted to the
hospital. Table 2 summarizes the TCD and SPECT findings in both groups.

solved in 2 days and 91%, in 2 weeks. Sixty-four perce
of initial MCA stenoses resolved in 2 weeks.
Absence of perfusion on SPECT (SPD) was detect
in 19 of 40 stroke patients and was not seen in the T
group (0/10 patients). Normal perfusion pattern
SPECT was more common in the TIA group (5/10
1/40, p = 0.0003).

Correlation with Cerebral Angiography: Cerebral angiography was performed in 16 patients within the first
3 days after onset. Two MCA occlusions, 2 MCA stenoses, 2 developed collaterals, and 1 normal result were
found in 7 stroke patients. Three MCA stenoses, 2 developed collaterals, and 4 normal results were found in
9 patients with TIAs. In all patients, cerebral angiography
was reviewed independently by a radiologist and the reports confirmed the TCD and SPECT findings. The numbers are too small to apply statistical methods to estimate
the sensitivity and specificity.

The Association Between TCD and SPECT Patterns: TC

and SPECT results (Table 3) were compared. A stro
correlation was obtained in cases of initial TCD occl
sions and SPDs but there was no correlation betwe
results of interim and final TCD studies to the first a
second SPECT studies.
The Association Between TCD and SPECT Patterns a
Neurological Deficit Scoring: Neurological deficit scar

(on the CNS) on admission and 2 weeks later were co
pared between 19 patients with SPD/TCD occlusions a
those with other strokes (n = 23) (Fig 2). All nonoccl
sive TCD and SPECT patterns were associated with
better short-term outcome; there was no significant d
ference between them.

Presumed Middle Cerebral Artery Occlusion and Severe
Perfusion Deficits: The occlusive TCD pattern was com-

mon in the stroke group (19/42 patients) and was not
seen in the TIA group (0/18 patients). A normal TCD
pattern was prevalent in the TIA group (9/18 vs 8/42, p
= 0.02).
Increasing numbers of stenotic, collateral, and normal
patterns reflected spontaneous resolution of MCA occlusion on TCD (n = 19). The majority of initial occlusions
(n = 13) transformed into stenoses within the subacute
phase. Sixty-eight percent of initial MCA occlusions re-

Comparison of CT, SPECT, and TCD: The volume of tt
ischemic lesion was measured on later CT scans (>
days). All available CT scans in the TIA group (n = 1•
were negative. Stroke patients were divided into tVI

Table 3. TCD and SPECT Patterns of Perfusion in the Stroke Group
Initial TCD /First SPECT
N

N

c
s
0

H

2
2

5

L

SPD

5

2
4

19

N

Interim TCD /First SPECT
H
L
SPD

3
3
3

5

2
4

2
I
10

6

N

Final TCD/Second SPECT
H
L
SP

5
3
3

7

3
8

2

TCD: N = normal; C = collateral; S = stenotic; 0 = occlusive. SPECT: N = normal; H = high (including mixed); L = low; SPD = seve11
perfusion deficit (absence of perfusion).

Alexandrov et al: Cerebral Perfusion Index

21

lesion
volume,
em'

240

TIA: (n=14)
no lesions on CT-scans

Table 5. Cerebral Perfusion Index (CPI) Predicts
Short-Term Outcome (2 Weeks)

200
Non-occlusions (NO):
19+/-21 em' (n=20)

150
100

Occlusions (0):
147 +j-87 cm 3 (n=17)

p<0.0001

Fig 3. Mean volume of ischemic lesion in transient ischemic
attack (TIA) and stroke subgroups.

groups: those with occlusions (n = 17) and those with
other patterns (n = 20). Mean infarction volumes were
compared between these groups (Fig 3). In 15 patients
the first SPECT study was done after negative findings
on the admission CT and prior to the second positive
CT study (see Case Reports). In 12, SPECT showed perfusion defects corresponding to the lesions (in 3 patients
hyperperfusion was found and it was difficult to compare
images).
Forty-two
strokes were subdivided etiologically into four groups according to the following criteria:
Hyperperfusion in Cardioembolic Stroke:

1. Cardioembolic-22 patients-high TES score, e.g.,

atrial fibrillation
2. Atherothrombotic-5 patients-more than 70% stenosis of ipsilateral internal carotid artery (ICA) (carotid duplex)
3. Lacunar-3 patients-clinical and CT criteria
4. "Cryptogenic"-12 patients-no obvious cause was
found
In the TIA group, 6 TIAs were cardioembolic; 5, atherothrombotic; and 7, of unrevealed cause.
The distribution of SPECT types of perfusion versus
etiology was analyzed, and there was a tendency for hyperfixation ofHMPAO to be more common in cardioembolic versus other types of stroke and TIAs-14 high and

CPI•

15-20
6-12
1-5

CNS at 2
Weeks

n

111.5 ± 7.4
85.3 ± 32.5
50.2 ± 34.7

10
19
22

p
O.Ql

0.001

•CPI: 15-20 = good; 6-12 = medium; 1-5 = poor. CNS =
Canadian Neurological Scale.

mixed versus 23 low and normal patterns, compared to
8 versus 36 in the nonembolic group (n = 81 )-but it
was not significant.
CPI as a Predictor of Short-Term Outcome: Recovery of
the neurological deficit was subdivided into the following
groups:

1. Complete (with CNS score at 2 weeks-115)

2. Good (;;:;.90-115)
3. Partial (;;:;.50-90)
4. Poor (<50)
The mean values of CNS scores on admission and the
index derived from the first SPECT and initial TCD in
each group of patients above were compared (Table 4).
There was no correlation between CNS scores at 2 weeks
and CPis derived from other TCD and SPECT results.
Furthermore, the three grades ofCPI (1-5, 6-12, 15-20)
were compared to the short-term outcomes (mean CNS
scores at 2 weeks after onset), and a strong association
between them and different degrees of the recovery was
found (Table 5).
Case Reports
Patient 1: An 82-year-old woman with a history of cardiac disease was admitted within the first 2 hours after
sudden onset ofleft-sided hemiplegia. The CNS score on
admission was 45. The TES indicated a high risk of cardioembolism (atrial fibrillation). An occlusive TCD pattern (asonic right MCA and anterior cerebral artery
[ACA], and a flow in right ICA syphon) was detected. CT
demonstrated negative findings at 3 hours after onset.
SPECT was performed 12 hours after onset and showed

Table 4. Predicting Value of Neurological Deficit Scoring on Admission and Cerebral Perfusion Index (CPI)
Derived from the First SPECT and Initial TCD
Predictors

Recovery at 2
Weeks (CNS)

Initial CNS

n

Complete (115)
Good (;;;,.90-115)

75.4 ± 22.7
84.2 ± 12.2

22

Partial (;;;,.50-90)

68.2 ± 11.5

13
12

Poor (<50)

22.1 ±

13

8.9

CNS = Canadian Neurological Scale; NS = not significant.

212

Journal of Neuroimaging Vol 3 No 4 October 1993

p
NS
0.002
<0.0001

CPI

n

13.2 ± 5.7

12

8.3 ± 3.8
4.1 ± 3.0

13
12

1.4 ± 1.4

13

p
0.01
0.005
0.007

TCD

SPECT

First 2 hours

First 12 hours

100

First 12 hours

90

0

R

L

R

L

Occlusive TCD pattern and severe perfusion deficit on SPECT at 2 to 12 hours after onset of hemiplegia (see Case Report 1).
NF = not found.

Fig 4.

an absence of perfusion in the entire right MCA and ACA
territories. The initial CPI was 1. TCD showed persistence of MCA and ACA occlusions till the patient died
the next day (Fig 4). Cerebral infarction in the right MCA
and ACA territories was confirmed at autopsy.
Patient 2: A 73-year-old woman with a history of previous hemispheric TIAs was admitted within 14 hours
after onset of right-sided hemiparesis and global aphasia.
The CNS score on admission was 35. The TES indicated
a low risk of cardioembolism. The initial and interim
TCD studies showed an occlusive pattern. CT demonstrated negative findings at 16 hours after onset. SPECT
was performed at 24 hours and showed an absence of
perfusion in the left MCA and ACA territories. The initial
CPI was 1. Repeat CT scan was done on the fourth day
and revealed infarction in the corresponding territory
(infarction volume, 112.9 cm 3). The second SPECT study
was performed on the 14th day and prominent hyperfixation of the tracer was noted. The final TCD study
showed a patent left MCA and ACA (normal pattern)
(Fig 5). The CNS score at 2 weeks remained 35 and the
patient had a poor recovery, presumably due to late and
ineffective recanalization.
Discussion

This study showed that TCD and SPECT patterns in
stroke and TIA are quite different. Strokes are associated
with up to 50% TCDjSPECT occlusive patterns at 10

hours after onset, which correlates with angiographic
findings [11]. These patterns were never seen in the TIA
group where normal perfusion patterns are more common. Presumably it reflects the underlying pathogenesis.
In this study, only the initial TCD correlated with
SPECT findings (occlusive TCD pattern = absence of
perfusion on SPECT) while there was no significant correlation between interim and final TCD and SPECT studies in spite of them being done at the same time. Only
6 presumed occlusions persisted on TCD up to the time
of the first SPECT, when absence of perfusion was detected in 19 patients. So far, other TCD patterns displayed all possible SPECT patterns and vice versa. This
may be the reason why late or single TCD results are
difficult to correlate with subacute SPECT.
Luxury perfusion sometimes occurs in acute stroke
[ 16-18]. The present study showed that hyperfixation of
HMPAO on SPECT could be detected in up to 25% of
both stroke and TIA patients, which is slightly higher
than previously reported [ 17]. There is also a tendency
for the high and mixed HMPAO fixation SPECT patterns
to be more common in cardioembolic stroke. These
results correlate with the authors' previous findings in
two subacute SPECT studies involving 191 consecutive
stroke patients: High, particularly mixed patterns were
prevalent at day 3 and 2 weeks after onset in cardioembolic strokes [ 19].
Serial TCD and SPECT studies seem to relate to outcome. Occlusive patterns were found in all types of isAlexandrov et al: Cerebral Perfusion Index

213

0

TCD
Day 14

CT
Day 4

TCD
First 14 hours

46

80

SPECT

SPECT

First 24 hours

Day 14

42

Fig 5. TCD and SPECT patterns within 2 weeks of onset of hemiparesis and global aphasia associated with poor recovery. CT
demonstrates left middle cerebral artery infarction (see Case Report 2). NF = not found.

chemic stroke except lacunar and were associated with
the poorest short-term outcome. High and low SPECT
patterns as well as normal, collateral, or stenotic TCD
patterns were associated with better short-term outcome
compared to the recovery of patients in the occlusive
group. Presumably this represents early recanalization.
CPI values of l through 5 are presumably associated
with arterial occlusions and medium values (6-12), with
various combinations of reduced perfusion due to stenosed arteries or poor collaterals, possibly with low metabolism. High values of CPI (15-20) were associated
with normal or sufficient collateral perfusion and possibly
with normal or high metabolism. The predictive value
of CPI derived from the initial TCD and first SPECT
214

Journal of Neuroimaging Vol 3 No 4 October 1993

study was better than the initial clinical scoring (on the
CNS), and CPI provides information on a mechanism
of ischemia as well. Its predictive value in the immediate
phase of cerebral ischemia (within the first 6 hours) needs
to be evaluated prospectively. This index may prove clinically useful to identify the target group of patients with
maximum therapeutic response to thrombolysis in trials
designed without cerebral angiography.
In conclusion, TCD and SPECT have several promising advantages. Previous works have shown that TCD
is a reliable method in the diagnosis of MCA trunk lesions in acute stroke compared to angiography [12, 13],
while SPECT is more sensitive than CT in the acute
phase of stroke [5, 6]. Good predictive value was dem-

onstrated for both methods if performed acutely [8-1 0,
14]. The present study supports these findings, and serial
TCD and SPECT could contribute to the management
of thrombolytic therapy. Their combination (CPI) may
prove to be a safe, fast, and reliable noninvasive substitute for angiography.
This work is supported by Sunnybrook Trust for Medical Research Foundation. Dr Alexandrov is a Postgraduate Research
Fellow from the Institute of Neurology, Academy of Medical
Sciences, Moscow, Russia.
The authors are grateful to Sandra E. Black, MD, FRCP(C),
Beverly Bowyer, RN, and Philippa Johnston, BA, for their invaluable help and assistance.
References

I. Norris JW. Drug treatment of acute cerebral infarction.
In: Krigelstein J, ed. Proceedings of the Second International Symposium on Pharmacology of Cerebral Ischaemia. Stuttgart: Wissenschatfliche Verlagsgesellschaft,
1989:439
2. Hennerici M, Hacke W, von Kummer R, et a!. Intravenous tissue plasminogen activator for the treatment of
acute thromboembolic ischemia. Cerebrovasc Dis
199l;l(suppl 1):124-128
3. Brott TG, Clarke Haley E Jr, Levy D, eta!. Urgent therapy for stroke: Part I. Pilot study of tissue plasminogen
activator administered within 90 minutes. Stroke 1992;
23:632-641
4. del Zoppo GJ, Poeck K, Pessin MS, eta!. Recombinant
tissue plasminogen activator in acute thrombotic and
embolic stroke. Ann Neurol 1992;32:78-86
5. Yeh SH, Liu RS, Hu HH, et a!. Brain SPECT imaging
with 99Tc-hexamethylpropyleneamine oxime in the early
detection of cerebral infarction: comparison with transmission computed tomography. Nucl Med Commun
1986;7:873
6. Feldmann M, Voth E, Henze Th, et a!. Significance of
HMPAO-SPECT in the early diagnosis and followup of
acute cerebral ischemia-comparison to CCT. Psychiatry
Res 1989;29:457-458
7. Mountz JM, Modell JG, Foster ML, eta!. Prognostication of recovery following stroke using the comparison of
CT and technetium-99m HM-PAO SPECT. J Nucl Med
1990;31:61-66
8. Limburg M, van Royen EA, Hijdra A, Verbeeten B Jr.
r-CBF-SPECT in brain infarction: when does it predict
outcome? J Nucl Med 1991;32:382-387
9. Kushner MJ, Zanette EM, Bastianello S, eta!. Transcranial Doppler in acute hemispheric brain infarction. Neurology 1991;41:109-113
10. Giubilei F, Lenzi GL, Di Piero V, et a!. Predictive value
of brain perfusion single-photon emission computed tomography in acute ischemic stroke. Stroke 1990;21:895900
11. Fieschi C, Argentino C, Lenzi GL, et a!. Clinical and
instrumental evaluation of patients with ischemic stroke
within the six hours. J Neurol Sci 1989;91:311-322
12. Zanette EM, Fieschi C, Bozzao L, et a!. Comparison of
cerebral angiography and transcranial Doppler sonography in acute stroke. Stroke 1989;20:899-903
13. Kaps M, Damian MS, TeschendorfU, DorndorfW.
Transcranial Doppler ultrasound findings in middle cerebral artery occlusion. Stroke 1990;21:532-537
14. Cote R, Hachinski VC, Shurvell BL, et a!. The Canadian

Neurological Scale: a preliminary study in acute stroke.
Stroke 1986;17:731-737
15. Norris JW, Zhu CZ. Transcranial Doppler: a new test fo
cerebral embolism? In: Yao JST, Pearce WH, eds. Tech
nologies in vascular surgery. Philadelphia: Saunders,
1992:166-173
16. De Roo M, Mortelmans L, Devos P, et a!. Clinical expe
rience with Tc-99m HMPAO high resolution SPECT of
the brain in patients with cerebrovascular accidents. Eu
J Nucl Med 1989;15:9-15
17. Moretti JL, Gilles D, Cinotti L, eta!. "Luxury perfusion" with 99m-Tc-HMPAO and 123-1-lmp SPECT imaging during the subacute phase of stroke. Eur J Nucl
Med 1990;16:17-22
18. Leonard J-P, Nowotnik DP, Neirinckx R. Technetium99m-d,l-HM-PAO: a new radiopharmaceutical for imag
ing regional brain perfusion using SPECT-a compariso
with iodine-123 HIPDM. J Nucl Med 1986;27:1819-18
19. Alexandrov AV, Pirisi A, Black SE, eta!. Value of single
photon emission computed tomography in acute stroke.
Ann Neurol 1992;32:237 (Abstract)

8th International Symposium on Cerebral
Hemodynamics: Transcranial Doppler, Cerebral
Blood Flow, and Other Modalities
Date: February 14-17, 1994
Place: Hyatt Regency La Jolla at Aventine, San
Diego, CA
Contact: Arlene Wiedel
Conference Coordinator
Institute of Applied Physiology
and Medicine
701 16th Avenue
Seattle, WA 98122
Tel: (206) 553-7330; Fax: (206) 553-1717

Alexandrov et al: Cerebral Perfusion Index

215