Daniel Barbara

K. Kido, MD M. Rothenberg,

Christopher Cox, PhD MPA #{149} Paul D. Woolf,

Traumatic Predictive The

computed

scans

from

W. Hamill,

#{149}

MD

Brain Injuries: Usefulness of CT’

tomographic 72 patients

E

(CT)

with

in neurologic with the Glasgow

function Coma

imaging

MERGENCY

studies

are

rou-

tineby obtained as part of the mitial evaluation process in patients with acute head trauma to determine whether a surgically correctable be-

traumatic

brain injury were reviewed to determine whether a specific type, location, or size of lesion correlated with changes sessed

Robert MD

#{149}

(asScale

sion

[GCS]), patient outcome (assessed with the Glasgow Outcome Scale [GOSI), or catecholamine levels. The

exists.

Computed

tomography

(CT) is currently choice because readily available,

the procedure it is faster and and it more

accommodates

emergency

of more easily

equipment

lesions were classified as focal or diffuse. GOS changed as a function of lesion size (P = .00004) in the 48 patients with focal hemorrhages, regardless of whether the lesions were

than magnetic resonance (MR) imaging (i-3). In addition, CT can enable detection of blood during the acute phase, while MR imaging with noutine sequences may not be able to do

intra- or extraaxial, tients with normal with lesions larger

so. The

had

a twofold

outcome

greater

than

lesions

and in the i9 paCT scans. Patients than 4,iOO mm3

(iOO%

risk

with

vs 50%).

Patients

smaller with

normal CT scans were significantly more likely to have mild neurological dysfunction or none than patients with abnormal CT scans (P = .03), but lesion location, skull fracture, and pineal shift were not significant predictors of GCS or GOS scores. A positive lesion

relationship size and both

existed plasma

between

(P

lesion

between norepi-

size

and

the

GCS

10). Index

terms:

orrhage, Trauma,

Radiology

I From (P.D.W.)

NY. cepted

Brain,

10.434 iO.434

1992;

the and

Received

CT,

10.i2ii

#{149}Brain,

injuries,

November

cohol Abuse and ogy, Washington ‘ RSNA, 1992

hem-

iO.434

of Radiology of Biostatistics

8, 199i.

8, 1989;

Alcoholism. University

(D.K.K., (CC.),

revision

Supported

Address Medical

in part

B.M.R.), University

requested by

reprint Center,

observed

patient’s

on CT course

and

has

been

studied

with

CT

and

Norepinephrine

levels

are

higher

in patients with more severe brain injury (graded with the Glasgow Coma Scale [GCS] [ii]) than in patients with mild or moderate injury and may enable prediction of outcome in patients with severe injury. Other researchers have found that

i82:777-78i

Departments the Division

October

Brain,

#{149}

and

stem

.02).

=

lesions

scans

grant

stimulus

for

the

observed

sympathetic

activation remains The pathologic

unknown. and physiologic

sons

catecholamine

for

elevated

realevels

in head trauma are not entirely clear. One way to examine this phenomenon is to assess the location and extent of brain injury with neuroimaging techniques and ascertain whether sympathetic

nervous

system

tion, identified with plasma cholamine levels, is associated specific

in-

MR imaging (7,8). Correlative studies of diseased tissue and anatomic perturbations shown by CT and their relationship to metabolic and systemic derangements that attend brain injury remain to be fully elaborated. We have previously demonstrated an excellent correlation between serum catecholamine levels and the severity of neurologic dysfunction (9,

nephrine and epinephrine levels (P < .02); a significant relationship existed score

tracraniab

between

prognosis have previously been examined (4-6). More recently, the relationship between level of arousal and downward displacement or herniation of brain structures near the diencephabon, tentonum, and upper brain

of a poor

patients

relationships

blood pressure and heart rate are positively correlated with serum catecholamine levels (12). However, the

locus

or specific

activa-

catewith

a

boci of brain

injury or with a diffuse overall insult. In our study, we attempted to determine whether a specific type (intraon extraaxial), site (frontal, parietab, temporal, or occipital lobe; basal ganglia; brain stem; or cerebellum), or size of lesion correlates with changes in neurobogic function, patient outcome, or catecholamine levels. PATIENTS

AND

METHODS

Seventy-two patients matic brain injury were study

as part

of the

who had trauenrolled in this

Head

Trauma

Project

at the University of Rochester Medical Center, Rochester, NY. These patients, aged 17-95 years (median age, 27 years), had minimal systemic injury (ie, they had no intraabdominal or intrathoracic injury, although some had bone fractures). Informed consent was obtained from all patients or from the responsible relatives.

CT scans

were

obtained

with

a commer-

cially available scanner ical Systems, Milwaukee)

(GE 8800; GE Medshortly after ad-

mission

Toward

to the

hospital.

the

end

of

the study, a few scans were obtained with a GE 9800 scanner. All scanning was performed 15#{176} to the Reid baseline by use of the scout view. All scans were contiguous, i-cm-thick sections from the foramen

Neurology (R.W.H.), and Medicine of Rochester Medical Center, Rochester,

December ROi-AA-07066

requests to D.K.K., 5i0 S Kingshighway

14;

final from

revision the

National

received Institute

and

acof Al-

Mallinckrodt Institute of RadiolBlvd, St Louis, MO 63110.

Abbreviations: GOS = Glasgow positron emission

GCS

=

Glasgow

Outcome

Coma

Scale, PET

Scale, =

tomography.

777

to the

magnum rial

was

operated

on

the

on the 9800 All CT

kVp

and

120

8800

unit

or for

were

read

(D.K.K.)

according

mA

for

9.6

2 seconds

were

classified

lesions

were

as focal

extraaxial.

The

summarized and

lesions

or diffuse.

Focal

as intra-

were

1. The

focal

areas

of

also

cause Most

of the small number of these cases. CT scans were obtained shortly after

calculated on

peared

the

diameter the

was

the

largest

was

was

(Fig

measured

the

chosen.

The

and

considered the

perpendicular

to

the

entire

in between mm. one

Whenever section,

6 mm. height

pineal

whom assigned

CT

assigned

which

ellipsoid. The rectly measured

size

a value

was

in millimeters.

scans

appeared

sented

in Figures

2 and

days).

did

not

samples

die

was

were

within

48 hours

catecholamine

levels

technique

15 days

obtained

with

on

studies

only to be

mdiof

Patients

in

be repre-

3.

demonstrated

levels

varied

mean

without

that

less

than

any

25%

around

apparent

diurnal

and

7.7% 6.9%

and

16.8%,

and

14.4%. between levels

The relationships (Ii) catecholamine

GOS

scores

were

and

studied

for

the

ventricular

system

space was also ium was checked

The

Finally, fractures.

of the head

outcome

recorded within

the

or subarachnoid

noted. for

severity

patient’s

were of blood

were

the

injury

calvar-

and

determined

the with

standard methods. The GCS of Teasdale and Jennett (1 1), which incorporates measures of the best motor and verbal responses and eye opening, was used to assess the neurologic function (Table 2). Neurologic

abnormalities

were

scored

and

therefore

analysis.

778

6 months later were considered The

Radiology

#{149}

were median

(13). The 6-month more reliable

used length

throughout of stay

methods

with

levels

analysis

(14).

were

to make

case

log-transformed them

more

the for

12 (17)

3 (4) 9 (i2) 4 (6)

16 (22) 4 (6) 2 (3)

blood Normal

of the scores, were cate-

Patients

Intraaxial and extraaxial Intraaxial larger than extraaxial Extraaxial larger than intraaxial Diffuse (excluded from analysis) Diffuse edema Ventricular or subarachnoid

appropriate

In the

of Lesion

Multiple

size and

3 (4) 19(26)

Total for

72

Note-Numbers ages.

in parentheses

are percent-

normally

distributed. We also used the natural logarithm of lesion size as a predictor, because the volume varied over several orders of magnitude. A log-linear regression model, which assumes that the effect of the independent variable is multiplicative rather than additive, was used for analysis of GOS scores. Statistical significance was assessed with likelihood-ratio x2 tests on the basis of the Poisson distribution (i4).

as

follows: 3-4, severe; 5-7, marked; 8-10, moderate; and 11-15, mild. The GOS was used to classify the patient’s status at the time of discharge or transfer from the hospital and numbers

regression

of

Single

epi-

(a) lesion and GCS

cholamine

fossa

presence

varia-

tion, the mean catecholamine level in each patient was considered representative (10). The respective intraand interassay coefficients of variation for norepinephrine were nephnine,

by Type

Focal Intraaxial Single Multiple Extraaxial

the

ganglia

posterior

hemor-

indicate the arrow) and width

No. of

catecholamine

catecholamine levels and the GCS ordinary linear regression methods used. As in previous studies (9,10),

The

intraaxial

of Patients

Type

Arprevious

location of each lesion was identifled according to the side of the head in which it occurred, as well as the lobe where it was centered. Lesions in the basal and

(a) Single

of injury to measure by use of radioenzycommercially avail-

of 10

The

separately.

hemorrhages.

Table i Distribution Lesion

(range,

twice

able reagents (Cat-A-Kit; Amersham, lington Heights, Ill). Because our

also shift

could

who

matic

ones

normal were size (0.3 of a

lesion scores

those 3-184 daily

as an

was the

gland

voxel),

and extraaxial

rhage in the left frontal lobe. Two-headed arrow indicates length; arrowheads width. (b) Two right-sided extraaxial hemorrhages. The length (two-headed (arrowheads) where the larger lesion was measured are shown.

Blood

and volume

modeled

of a lesion by recording

a nominal so that their

of intra-

length

the

the length, width, used to estimate the

of the lesion,

the

while

a lesion was seen its height was estimated

Fourth, were

Measurement

the greatest.

section),

were

1.

width

Third, the height was calculated by adding together the number of sections on which the hemorrhage appeared. The thickness of the top and bottom sections was assigned a value of 8 mm (because it was not clear whether the lesion extended through

D.

Figure

largest

to the

appeared

a.

CT ap-

I). Second,

the width

be-

was

First,

hemorrhage

measured

length

where

steps.

which

analysis

hemorrhage

in several

section

be

of each

this

obtained lesions

were

admission. The volume

from

of the

because

the CT scan with diffuse

Patients

excluded

are

with separately,

not analyzed on

or

division

edema associated were measured

their infrequency at admission.

of

revealed The

of this

in Table

infarction hemorrhages

scan

classified

results

one

categorized

normal.

further

by

were

the

or appeared

they

was

blindly

and

to whether

lesions

mate-

scanner

unit.

scans

authors

but

No contrast The

at 120

seconds

the

vertex.

administered.

the injuries nies

were

[27%]),

pedestrian

[i4%]),

assaults

ing

accidents

were

study

subjects,

(31.3

the by

25

=

=

inju-

13

= (ii

=

7

[6%1), and ski-

Ethanol in 23 of the

2 [4%]).

available greater

3

other

(n

falls accidents

(n (ii

levels

nob levels

RESULTS

(48%); caused

10 of whom

than

had

etha-

100 mg/dL

mmol/L).

Thirty-five of the 48 patients with focal intracranial hemorrhages were men; 15 of the 19 patients in whom

The normal traaxial

median age of patients with CT scans and those with inbrain injury only was 22

CT

years; rhages

in those with extnaaxial hemoronly, 36 years; and in those

Motor

scans

appeared

vehicle

normal

accidents

were

caused

men.

23 of

March

i992

B

D

A

MA

hA

outcomes. A good outcome was defined as a GOS scone of good recovery or moderate disability; a poor outcome, as a GOS score of severe dis-

A

!

A

AA

C

A

Pv

ability, AA

A

A

A

A

A

A

A

A

A

In-

C AMA

A

I SD

A

A

B

A

(0 A

MD

MB

A

F

I0

C

m

B

A

A

A

A

A

AA

A

AA

AA

AA

AA

10

I

100

1000

10000

100000

Size (mm’)

_ .

G

_I

A

1

10

100

1000

10000

100000

Figure 3. GCS size (expressed

score

score as a function in cubic millimeters).

of lesion

Figure

2.

Glasgow

Outcome

score as a function in cubic millimeters).

ure 3, each

Scale

of lesion In this

(GOS)

mab CT scans,

size (expressed figure and Fig-

letter

indicates the number of (eg, A = 1, B = 2, C [not 3). D = death, G = good recovery,

data points shown] = MD = moderate

disability

(disabled

pendent), PV = persistent vegetative SD = severe disability (conscious but

but indestate, dis-

abled).

Table 2 Glasgow

Scale Status

Eye opening

Nil Best motor

(E)

4 3 2 response

(M)

6 5 4 3 2

Nil

1

response

Oriented Confused

from score

with sions,

both intra35.5 years.

with

normal

reference

and extraaxial None of the

CT scans

than age 50 years. In the 48 patients

Volume

lesions

182

was

and

3-5.

lepatients

older

than

14 of the 48 pawere older

with

Number

#{149}

6. =

the

3

focal

in-

i9 with

shown

lesion

norepinephrine

2 1

(E + M + V)

age 50 years, whereas tients with focal lesions

tracranial

which

3

Nil

results

of a regres-

6-month

GOS

versus

extraaxial)

two such lesions existed. both GCS scores and norlevels have previously

to be predictors

scores (9,10), we performed regression analysis of GOS

5 4

conversation

Source-Adapted Note-Coma

been

(V)

Inappropriate words Incomprehensible sounds

(intraaxial

cause only Because epinephrine

1

Obeys Localizes Withdraws Abnormal flexion Extensor response

the of the

failed to reveal any significant effects of these independent parameters on GOS scores. No attempt was made to apply GOS scores to lesions located in the infratentorial compartments be-

Score

Spontaneous Tospeech Topain

analysis

score on the natural logarithm of lesion size revealed that the GOS score changed as a function of lesion size. The curved line in Figure 2 is the predicted mean score from the regression. The slope coefficient of this curve was statistically significant (P = .00004). Study of the importance of the location and type of individual lesion

Coma

Neurologic

Verbal

sion

nor-

size,

GCS level

score, were

of GOS a multiple scores in

and used

as

predictors. Even when adjusted for these two important, additional vanables, the logarithm of lesion size memained a predictor of GOS score (P = Oil). In this analysis, norepinephrine was also significantly associated with outcome (P = .054), while the GCS score was not related to the GOS score (P = .21), after both bogarithms were controlled for lesion size. This suggests that catecholamine 1evels contain information that is not included in data on the size of the besion.

We further explored the effects of lesion size on prognosis by grouping patients according to good or poor

on

vegetative

state,

admission

and

both

levels,

which

was

cholamine ousby

SIZE (mm’)

persistent

or

death. All patients with lesion volumes larger than 4,100 mm3 were included in the poor-outcome group, compared with only 50% of those with lesion volumes less than 4,iOO mm3. Logistic regression analysis mevealed that the probability of a poor outcome increased with increasing lesion size (P = .000001). This melationship persisted (P = .0003) when adjustment was made for the admission GCS score and norepinephnine level, which was also a significant independent predictor of outcome (P = .0009). The relationship between GCS

reported

by

Woolf

cate-

previ-

et ab (9) and

Hamill et al (10), was also present in this study (P < .0001). In addition, using bivariate regression analysis with lesion size and GCS score on admission as independent variables, we found a positive relationship between lesion size and both catechobamine bevels (P < .02). A significant relationship also existed (P = .02) between lesion size and GCS score on admission: Larger lesions enabled prediction of lower GCS scores (Fig 3). Furthen study of the effects of lesion type, location, presence of pineab shift, or skull fracture failed to demonstrate an association between these parameters and either GCS or GOS score. Patients with normal CT scans at the time of admission were significantly more likely to have no on mild neurologic dysfunction (GCS scores of il-i5) than those with abnormal CT scans (x2 = 4.72, df = 1, P = .03). The difference between patients with normal and those with abnormal CT scans was also present in their outcomes. Ninety-four percent of the group with normal CT scans (n = 16) had either good recoveries or moderate disabilities versus 49% of those who had intracranial hemorrhage (n = 22) (x2 = iO.64, df = 1, P = .OOi) (Table 3). However, a normal CT scan did not necessarily imply that the patient recovered fully; seven of these patients (41%) remained moderately or severely disabled. Similar data have also been reported by other groups (15). Nevertheless, no patient in the group with normal CT scans became persistently vegetative or died, compared with ii patients (24%) with focal inRadiology

#{149} 779

Table 3 GOS after

Months

6

and Mean

Lesion

Size Focal

bntracra

nial

Hemorrhages

Values

Intraaxial

GOS

Score

Extraaxial

No.

Good

No.

Size*

6

289

Size*

4

disabled

Severely

2

disabled

Persistent

vegetative

state

0

Death

2

1 4

per

lesion

category

14

0 3

#{149}Mean size of lesion

tracranial df= i,P

in cubic

12

3,303

(x2

hemorrhages

=

plus

or minus

associated

with

a worse

GCS

scone

and prognosis (5), and intracranial hematomas have been associated more frequently than diffuse brain injury with neurologic deterioration (6).

Although

firm

we

a specific

suggest

were

unable

relationship

pathologic

process, that,

to conto

our

of the

data

parameters

a single

clearly exam-

med. lesion size is most critical. This may not be surprising because increasing lesion size should compress and distort vital structures in the diencephabon and upper brain stem A number

of basic have

been

neaddressed

in

view of recent CT and MR findings. For instance, the exact sequence of events associated with brain injury and the level of consciousness is no longer clear. Contrary to data derived from clinical-pathologic studies, which suggested of consciousness with hemispheric

related appears 780

that the depression observed in patients mass lesions was

to transtentonial that the level Radiology

#{149}

henniation, it of arousal con-

4

19

(17,604) 9,654

2

best

with

the

distortion

study

of the

confirmed

establishment

omy of the area posterior fossa.

(7).

the

and dipre-

of a clear

of the tentorium and Subsequent studies

will require finer imaging to resolve the possibility of relevant lesions in this area. Because of its greater sensitivity in detection of lesions in this area and because of its capacity to image in three dimensions, MR imaging to be the

best

modality

for

addressing this issue (18). In addition, readers of CT scans may underestimate lesions (hemorrhages) at sites distant

from

the

vital

structures

in the

brain stem and diencephabon. Our previous research indicated that serum catecholamine levels GCS

study ship

score

scores

and

enable

(9,iO).

The

re-

prediction

current

shows that a similar relationdoes exist between (a) hemor-

rhage cholamine

size

Size*

15

‘I’

10

492

(140) 3,816

7

6

(2,953) 4,233 (1,640) 9,547 (9,302) 18,316

12 2 9

1 0 0

(8,711) 5,974

45

Scans

17

(2,045)

exduded from this analysis, as well as five patients in whom 6-month was used for analysis. In columns 2, 4, 6, and 8, No. = number of pa-

relationship between clinical status and brain stem anatomy. Our own CT studies do not provide the resolution required to depict the discrete anat-

appears

No. of Patients with Normal

(in parentheses).

MR

cluding

No.

(4,422)

secondary to lateral displaceof supratentorial structures

of GOS

pathophysiobogic

4

(339) 7,549 (3,396) 9,547 (9,302) 31,652

mole of horizontal brain distortion indicated that vertical brain stem mensions vary substantially (17),

flect

(7,17).

questions

error

A second

Our findings support and extend previous data in the neurosurgery literature that suggest CT scans may provide important prognostic data on neurobogic status and outcome (6,16). We demonstrated a clear relationship between the size of intra- on extraaxiab hemorrhage and outcome. The presence of subdural hematomas has been

1,021

3,263

standard

brain ment

DISCUSSION

4

or subarachnoid blood were was found, the largest lesion

relates

5.05,

.02).

=

1,728 NA

(2,132)

edema or ventricular more than one lesion

millimeters

804

(333) 876

(7,981)

(1,731) Note-Five patients who had diffuse follow-up was not available. Whenever tients. NA = not available.

5

(397) NA NA 10,573

(2,959) Values

Size*

(252)

10,740 (10,727) 4,128 (3,282) NA NA 3,261

4

No.

406

(112) Moderately

per GOS Category for All Focal Hemorrhages

Intra- and Extraaxial

and levels.

(b) GCS The

score presence

and

cateof

lesions in the limbic system, dorsolaterab frontal cortex, hypothalamus, and brain stem-where they might activate the sympathetic nervous system and cause catecholamine releasemight be established with other imaging modalities such as MR imaging or positron emission tomography (PET),

which are more sensitive than CT in detection of both morphologic and metabolic lesions. MR images have shown shear injuries and contusions in patients in whom the corresponding CT scans are normal (2,19-21). MR imaging was superior to CT in depicting 23 of 41 extracerebral lesions and in visualizing nonhemorrhagic contusions in 15 of 21 lesions (1). In another study, 19 brain stem lesions were identified

with

CT,

while

48 lesions

were present on the corresponding MR images (2i). MR imaging is thus ideal for demonstrating subtle lesions in the diencephabon and brain stem, where lesions might activate the sympathetic nervous system after brain trauma. PET and single photon emission CT have revealed both increased and decreased metabolism in areas where the CT scans and MR images are normal (22,23). Nevertheless, the use of MR imaging and PET will be restricted in patients with severe head trauma until the area around the MR imager can accommodate monitoring devices more readily and until PET can be performed more rapidly and bess expensively. How these modalities complement each another and CT thus remains to be determined. In the meantime, CT will continue to be the most widely used modality, especially for evaluation of severely injured patients whose condition is unstable. It is therefore

important

to understand

what information CT can provide. The capacity of CT to depict surgical lesions is well known. Furthermore, early studies of head trauma examMarch

1992

med

with

between

and

CT identified the

level

of consciousness

the frequency

findings fortunately tate the

of abnormal

CT

(4,18), but these studies undid not attempt to quantisize of the lesions. In contrast,

our study demonstrates tion between lesion specifically,

tient

a relationship

lesion

outcome.

an associasize (or, more volume)

Our

that the prognosis uabs with a lesion mm3. U

findings

and

6.

7.

indicate

8.

9.

References 1.

2.

3.

Snow RB, Zimmerman RD, Gandy SE, Deck MDF. Comparison of magnetic resonance imaging and computed tomography in the evaluation of head injury. Neurosurgery i986; 18:45-52. Zimmerman RA, Bilaniuk LT, Hackney DB, Goldberg HI, Grossman RI. Head injury: early results of comparing CT and highfield MR. AJNR 1986; 7:757-764. WilbergerJE, Deeb Z, Rothfus W. Magnetic resonance imaging in cases of severe head injury. Neurosurgery 1987; 20:571576.

4.

5.

French BN, Dublin AB. The value of computerized tomography in the management of i,000 consecutive head injuries. Surg Neurol i977; 7:171-183. Gennarelli TA, Spiebman GM, Langfitt TW, et al. Influence of the type of intracranial lesion on outcome from severe head injury: a multicenter study using a new clas-

Volume

182

#{149} Number

3

Ropper brain with

pa-

is poor for individlarger than 4,100

sification system. J Neurosurg 1982; 56:2632. Clifton GL, Grossman RG, Makela ME, Miner ME, Handel S, Sadhu V. Neurobogical course and correlated computerized tomography findings after severe closed head injury. J Neurosurg 1980; 52:611-624.

10.

ii.

12.

13.

14.

15.

displacement

of the

and level of consciousness an acute hemispheral mass.

AH.

Lateral

in patients N Engl

16.

17.

18.

Med 1986; 314:953-958. Stovring J. Descending

tentorial herniation: findings on computed tomography. Neuroradiology 1977; 14:101-105. Woolf PD, Hamill RW, Lee LA, Cox C, McDonald JV. The predictive value of catecholamines in assessing outcome in traumatic brain injury. J Neurosurg 1987; 66: 875-882. Hamill RW, Woolf PD, McDonald JV, Lee LA, Kelly M. Catecholamines predict outcome in traumatic brain injury. Ann Neu-

20.

rob 1987; 21:438-443.

21.

B. Assessment of consciousness: a practical scale. Lancet 1974; 2:81-83. Clifton GI, Ziegler MG, Grossman RG. Circulating catecholamines and sympathetic activity after head injury. Neurosurgery 1981; 8:10-14. Jennett B, Bond M. Assessment of outcomes after severe brain damage: a practical scale. Lancet 1975; 1:480-484. McCullagh P. NelderJA. Generalized linear models. London: Chapman & Hall, 1983; 127-133. Lobato RD. Sarabia R, Rivas JJ, et ab. Normal computerized tomography scans in severe head injuries: prognostic and clinicab management implications. J Neurosurg 1986; 65:784-789.

19.

Lobato

RD. Cordobes F, Rivas JJ, et al. Outcome from severe head injury related to the type of intracranial lesion: a computerized tomography study. J Neurosurg 1983; 59:762-774. Ropper AH. A preliminary MRI study of the geometry of brain displacement and level of consciousness with acute intracranial masses. Neurology 1989; 39:622-627. Merino-deVillasante J, Taveras JM. Cornputerized tomography (CT) in acute head trauma. AIR 1976; 126:765-778. HesselinkJR, Dowd CF. Healey ME, Flajek P, Baker LL, Luerssen TG. MR imaging of brain contusions: a comparative study with CT. AJNR 1988; 9:269-278. Levin HS, Amparo E, Eisenberg HM, et al. Magnetic resonance imaging and computerized tomography in relation to the neurobehavioral sequelae of mild and moderate head injuries. J Neurosurg 1987; 66: 706-713.

Gentry

Traumatic Radiology

Teasdale G, Jennett coma and impaired

22.

23.

LR, Godersky

JC, Thompson

brain stem injury: 1989; 171:177-187.

BH.

MR imaging.

Abdel-Dayem HM, Sadek SA, Kouris K, et al. Changes in cerebral perfusion after acute head injury: comparison of CT with Tc-99m I-IM-PAO SPECT. Radiology 1987; 165:221-226. Langfitt TW, Obrist WD, Alavi A, et al. Computerized tomography, magnetic resonance imaging, and positron emission tomography in the study of brain trauma: preliminary observations. J Neurosurg 1986; 64:760-767.

Radiology

781

#{149}