the effect of elevated intracranial pressure on the vibrational response of the ovine head

Annals of Biomedical Engineering, Vol. 23, pp. 720-727, 1995 00913-6964/95 $10.50 + .00 Printed in the USA. All rights reserved. Copyright �9 1995 Biomedical Engineering Society

The Effect of Elevated Intracranial Pressure on the Vibrational Response of the Ovine Head

MILAN STEVANOVIC,* GEORGE R. WODICKA,*'t JOE D. BOURLAND,~" GEORGE P. GRABER,~"

KIRK S. FOSTER,t GARY C. LANTZ,t~: WILLIS A. TACKER,t:~ and ALLEN CYMERMANw

*School of Electrical and Computer Engineering, tHillenbrand Biomedical Engineering Center, :~School of Veterinary Medicine, Purdue University, West Lafayette, IN, w Army Research Institute of Environmental Medicine, Natick, MA

Abstract--Although potentially fatal increases in intracranial pressure (ICP) can occur in a number of pathological conditions, there is no reliable and noninvasive procedure to detect ICP elevation and quantitatively monitor changes over time. In this experimental study, the relationships between ICP elevation and the vibrational response of the head were determined. An ovine animal model was employed in which incremental increases in ICP were elicited and directly measured through intraventricular cannulae. At each ICP increment, a vibration source elicited a flexural response of the animal's head that was measured at four locations on the skull using accelerometers. Spectral analysis of the responses showed changes in proportion to ICP change up to roughly 20 cm H20 (15 mm Hg) above normal; a clinically significant range. Both magnitude and phase changes at frequencies between 4 and 7 kHz correlated well (~/> 0.92) with ICP across the study group. These findings suggest that the vibrational response of the head can be used to monitor changes in ICP noninvasively.

Keywords--Intracranial pressure, Sound, Vibration, Sheep.

INTRODUCTION

Problem Statement

The brain is surrounded by cerebrospinal fluid that pro- vides a cushion against sudden movement as well as nour- ishment for brain cells. The pressure of the fluid, known as intracranial pressure (ICP), is carefully controlled by homeostatic mechanisms in the body. In pathological cir- cumstances such as traumatic head injury, stroke, and acute mountain sickness (6) elevated ICP poses a clinical problem which must be carefully managed to prevent se- vere brain damage or even death of the patient. Currently, the only methods for monitoring ICP are invasive and include the use of intraventricular catheters, subarachnoid

Acknowledgment--The authors thank John L. Semmlow for his en- couragement and helpful suggestions. This study was supported by the U.S. Army Medical Research and Development Command (DAMD17- 92-J-2025).

Address correspondence to G. R. Wodicka, Purdue University, 1285 Electrical Engineering Building, West Lafayette, IN 47907-1285, U.S.A.

Received 7Jun94, Revised 30Mar95, Accepted 29Jun95

or epidural pressure sensors, and the lumbar puncture (11). Although, these procedures are routinely performed by skilled physicians, they are not without potential com- plications, including infection. Thus, there is a clinical need for a noninvasive technique to reliably detect ICP elevation and monitor pressure changes over time.

A Potential Acoustic Solution

The analysis of sound transmitted through a structure or system is a commonly used engineering technique to assess underlying physical properties. The attributes of even highly heterogeneous systems such as the undersea envi- ronment can be determined by probing the system over a wide range of sonic frequencies and measuring the acoustic response. In the case of constrained structures, such as the human head, the response to sound (or equivalently, vibration) typically exhibits well-defined spectral and spatial patterns which are determined by the inherent geom- etry, weight, pressure, and losses in a predictable manner. Thus, it has been hypothesized that increases in ICP would be reflected in the overall vibrational properties of the head and could potentially be monitored noninvasively via response measurements.

Literature Review

In a preliminary study by Semmlow and Fisher (13) using two canine pups, the response of the head to low- level audible vibrations was observed to correlate with ICP elevation. In the protocol, the animals developed a chronically elevated ICP over a 6-week period due to initial autogenous blood injections into the cisterna magna, with a maximum ICP increase of roughly 11 cm H20 (8 mm Hg). An electro-mechanical solenoid was employed before and after increased ICP development to introduce vibrations into the heads of the animals, and the responses were measured on the opposite side of the skull using a contact accelerometer. Significant alterations in the initial timing, frequency content, and temporal decay of the response were noted after increased ICP development, and

720

Vibrational Monitoring of Increased Intracranial Pressure 721

the decay was observed to be a roughly linear function of ICP change. Although the small study group and the pos- sible effects of skull growth over the 6-week period com- plicate the interpretation of the results, the significant changes in the measured parameters suggest a sensitive relationship between noninvasive measurements and ICP.

The mechanical properties of the head have been studied extensively to understand the effects of the large forces and displacements associated with traumatic injury. In response to these large forces, the action of the head is highly nonlinear and significant structural damage can of- ten be observed. The response of the head to low-level acoustic vibrations, however, has been the topic of relatively few studies. To investigate the mechanisms of bone- conducted sound on hearing, the human skull (2) and head (1,2,4) have been vibrated using various techniques, and measurements of the response performed. Similar studies have also been performed on the human skull (10), cadaver head (3,7,14), and human head (3,8) as a prelude to further studies using larger forces.

Although each study employed a unique range of per- turbation forces and frequencies, measurement sites, and mechanical boundary conditions, some consistencies in the observations are evident. One such observation is that the head exhibits measurable resonance behavior over the frequency range between 100 and 2,000 cycles/sec (Hz). In this context, resonance can be loosely defined as the frequency at which a large response of the system occurs for a given input vibration. For various skulls, the first excitable (fundamental) resonance occurred at frequencies above 500 Hz, and the effects of the intracranial tissues was to decrease the observed resonance frequencies by roughly a factor of two. The effects of the extracranial tissues was to slightly decrease the resonance frequency even further and also to significantly increase the overall losses, or damping, of the response. It is clear from these measurements that the exact locations of both the transducers and the mechanical boundary (support) conditions on the head significantly affect the overall measured response. The former observation indicates that the head exhibits a flexural response (mode) to the vibrations since as the head flexes its measured response would exhibit strong spatial dependence. These flexural responses are also consistent with the strong dependence of the measurements on the mechanical support (or lack thereof) employed for the head in each study since the exact flexural mode of a structure is a strong function of the imposed boundary conditions.

Primary Goals of This Study

The principle objective of this experimental investiga- tion was to determine quantitatively the relationship be-

tween the vibrational properties of the head in the audible range and ICP elevation. The study incorporated an acute ovine model in which accurate measurements were made as a function of elevated ICP over a wide pressure range. Vibrational measurement and signal analysis techniques were employed to determine the parameters that were most sensitive and reliable correlates of ICP change. The results provide a framework upon which to develop a novel instrument that could be employed to noninvasively monitor ICP changes in both hospital and ambulatory care settings.

METHODS

Animal Surgery

An ovine animal model was employed to experimen- tally elevate ICP and measure the resulting changes in the vibrational properties of the head. This model has been successfully used in surgical studies of the cerebrospinal fluid (CSF) system (5) and was chosen because its skull size and shape facilitated the application of transducers to measure both ICP and the vibrational response. An acute protocol was used that mimicked the clinical syndrome found in acute mountain sickness and allowed for a wide range of generated ICP in a single animal.

A total of eight sheep (ovis aries) between 11 and 12 months of age and 55 and 65 kg body wt were studied. Complete vibrational and ICP data were obtained from four sheep (#4, 5, 6, 8), control data at normal ICP was taken over time in one of the sheep (#3) , and limited or no data was obtained from three sheep due to computer dif- ficulties (#1), intermittent catheter blockage (#2) , and hemorrhage (#7), during this complex experimental protocol. Each sheep was fasted for 24 hr with free-choice water before anesthesia induction. Anesthesia was induced with pentobarbital sodium (Anthony Products) 25 to 30 mg/kg, administered intravenously to effect, and then maintained with intravenous pentobarbital sodium and ox- ygen administered via an endotracheal tube. A ventilator assisted respiration. A right femoral arterial catheter was placed for direct continuous physiograph blood pressure monitoring. The sheep were placed in sternal recumbency. A midline cranial incision was made from the level of the medial canthi to 2.0 cm caudal to the external occipital protruberance. Soft tissues were subperiosteally elevated and retracted. A hole was drilled and tapped at the level of the internal occipital protruberance for screw mounting of the shaker head (Fig. 1). The external bone cortex at this level was contoured with a bone burr to ensure maximal contact between the shaker head and bone. Three accelerometers were mounted on the skull with screws placed in drilled and tapped holes located on the external occipital protruberance, and on each frontal bone. Two holes were also drilled and tapped for the mounting of a wire frame to

722 M. STEVANOVIC et al.

�9 Accelerometer 0 Catheter | Shaker/Impedance Head [ ] Skull Bone

FIGURE 1, Diagram of catheter and transducer locations on the ovine skull.

which the inflow and outflow lines attached to the ventricular catheters would be mounted. None of the six total holes that were drilled entered the cranial cavity. Trephine holes (0.5-cm diameter) were then made with a bone burr in the parietal bone 1.0 cm caudal to the coronal suture and 1.5 cm lateral to each side of the midline (12). The external dura mater remained intact. Teflon catheters (An- giocath, 19-gange, 11/4; Becton-Dickinson, Rutherford, N J) were modified by the addition of two side ports. The catheters with stylets were advanced until the respective lateral ventricle was penetrated about 12 to 15 mm ventral to the external dura. Removal of the stylet and return to the catheter hub of CSF confirmed placement of the catheters. The catheters were held in position with epoxy (5 Minute Epoxy Gel; Devcon Corporation, Danvers, MA). The wire frame was attached to the skull with screws. After attaching the catheter hubs to their respective inflow and outflow lines, the lines were secured to the frame with silk ties. This prevented the weight of the lines from con- tributing to kinking or dislodgment of the ventricular catheters.

Needle trocarization was performed as necessary during the protocol to relieve ruminal tympany. At the com- pletion of each experiment, while each sheep was under general anesthesia, potassium chloride was administered

and the ECG tracing monitored until all cardiac activity was stopped.

Measurement and Control of Intracranial Pressure

The tubing from the catheters leading to each lateral ventricle was connected to an infusion pump (model 994, Harvard Apparatus, Harvard, MA) and a pressure transducer (model CDX, Cobe Laboratories, Lakewood, CO). Initially, the infusion rate of fluid (Lactated Ringer's In- jection USP, Baxter Healthcare, McGaw Park, IL) was set extremely low, 0.00494 ml/min, so as to not elevate ICP significantly but yet insure catheter patency. The output of each pressure transducer was monitored using a chart re- corder (model Physiograph CPM, Narco Biosystems, Houston, TX) and both cardiac and respiratory variations as well as agreement between ICP values were noted. One of the catheters was chosen at random as the channel where the infusion rate would be increased to elevate ICP, and the ICP signal from the other catheter was connected to the computer for subsequent measurement. In this manner, ICP was elevated in a controlled manner via infusion into one lateral ventricle and measured via a catheter in the other ventricle. Infusion rates up to 0.494 ml/min were used during the protocol to allow vibrational measurements to be made at roughly 5 cm H20 (3.5 mm Hg) increments in ICP up to a maximum value of approximately 100 cm H20 (74 mm Hg).

Acoustic Measurements and Signal Processing

To elicit measurable vibrations of the head, noise with equal energy at frequencies between I00 and 10,000 Hz was generated (Model WSB-100, QuaTech, Akron, OH), amplified (Model NS B-2, Yamaha, San Diego, CA) and input to an electromechanical shaker (Model 203, Ling, Roystan, U.K.). The shaker was mounted in a free- standing frame that also served to immobilize the ovine head and thereby insure no bulk translation in response to the excitation. The shaker armature was connected to the proximal end of an impedance head (Model Z l l , Wilcoxen, Rockville, MD) that measures both force and acceleration, with the distal end mounted in the skull. The force and acceleration signals from the impedance head along with the signals from the three accelerometers mounted at the other skull locations (Model 303A, PCB, Buffalo, NY) were amplified using charge (Model 422D03, PCB) and voltage (model 483B07, PCB) ampli- fiers and then bandpass filtered (four-pole Butterworth type) with cutoff frequencies of 50 and 10,000 Hz. Each of the five acoustic signals (one force and four acceleration) were digitized with the ICP signal at a rate of 32,768 samples/sec (Model DAS-16F, Metrabyte, Taunton, MA) for a duration of 1 sec.

To simplify the interpretation of the signal processing


IMPEDANCE MAGNITUDES FROM FOUR SITES AT NORMAL ICP 105~ . . . . . . . . , . . . . . . . .

r [ - : Driving Point Impedance i! [ = Transfer Impedance 1 /~ ~. . . . . = Transfer Impedance 2 I [

} . . . . . . Transfer Impedance 3 /

~ "

"~ fO 3

. . . . . . . r . . . . . . . 10102 103 104

FREQUENCY [Hz]

IMPEDANCE PHASES FROM FOUR SITES AT NORMAL ICP 600 �9

: Driving Point Impedance / 500 = Transfer Impedance 1 / ,

- - = Transfer Impedance 2 / /

400i --' " - = Transfer Impedance 3 / /

3001

~. 200 r / , -/

E 100 - ' ~ '

103 FREQUENCY [Hz]

FIGURE 2. Absolute impedance (A) magnitude IZ,~f)l and (B) phase <;Z,~f) at the four measurements sites at normal ICP.

a Hanning function, transformed to the frequency domain, and then averaged to form the spectral estimate. Calibra- tion and mass data from each transducer were used to minimize transducer sensitivity differences and loading effects, respectively.

The absolute transfer impedance Zi(f) = FOO/Vi(J) to each output site was calculated and placed in magnitude ]Zi(D ] and phase 4s Zi(f) form. Zi(f) is a measure, as a function of frequency, of the relationship between the ap- plied force and the resulting velocity. To assess how Z;(f) changed with elevations in ICP, the relative impedance magnitude ] Zi( f, ICP)] and phase 4s ICP) with respect to the estimates at normal ICP were determined:

[Zir 1CP)[ IZi(f' ICP)t = IZ,r normal ICP) I

I Z(f, ICP) l = ~Zi( f , ICP)] - 4: ZiG, normal ICP).

Changes in the relative spectral magnitude or phase can be quantified in many ways such as calculating heights of spectral peaks or valleys. However, these measures can be difficult to extract accurately and therefore yield inconsis- tent results if the exact spectral shapes or peak locations change--as was observed to be the case for some of the ICP-induced spectral changes. Thus, after visual inspec- tion of the spectral data, the changes in each relative magnitude and phase spectrum were quantified by calculating the standard deviation of each over the 4- to 7-kHz frequency range where flexural responses were evident. In this context, since Zi( f, ICP) is relative to that at normal ICP, the standard deviation represents a global measure of the relative spectral energy in this band due to ICP elevation that is not sensitive to the exact spectral shape. The standard deviation was calculated as the square root of the mean square error of the spectral points relative to the

(Labview, National Instruments, Austin, TX), the four acceleration signals (i = 0, 1, 2, 3) were considered as outputs and processed relative to the force signal input. Initially, the coherence aft) between each output and the input was estimated and found to be very close to unity (>0.95) over the entire frequency, f, range of the excitation and measurements. This finding indicates a strong linear dependence of the acceleration response of the skull to low-amplitude audible vibrations as well as a high signal to background noise ratio of the measurements.

The output acceleration signals were each digitally in- tegrated (summed) to form the equivalent velocity. Spec- tral estimates of the velocity of each output Vi(f) and the input force F(f) were performed using the method pro- posed by Welch (9). Each time-domain digital signal was parsed into 14 epochs of 4,096 samples in duration with 50% overlap between successive epochs, windowed with

M A G N I T U D E OF R E L A T I V E D R I V I N G P O I N T I M P E D A N C E

lOP ELEVATION [cm H20]

. . . . . . . . . . . . . . . . . . . . . . . . . . i . . . . . . . . . . . . . . i ................ ................ i,ii. 1.o . . . . . . . . i . . . . . . . . . . . . . . . . . . . . . . . . . . .,.:: . . . . . . . . . . ii+

80 . . . . . . . . . . . . 'i ....... " ': : ~:i!

70 . . . . . . ~ " i. . . . . . . . . ~ :~]1

60 ~ . . . . . . . . ' . . . . . :. �9 . . . . . . . . . . . ~ :

so : ........ : . . . . , ' :i:i',

40 . . . . . ' i " . . . . . . . " ~11.

10 0 6000 8000 10000

0 2000 4 0 0 0 F R E Q U E N C Y [Hz]

FIGURE 3. R_epresentative relative driving point impedance magnitude IZ(f, ICP)I versus frequency and total ICP elevation.


iii A 8 I I I D 8

- - - - - 6 . . . . . . . . . . . i i i i .............. ................ . . . . . . . . . . . . . ............... .................. .............. ~ I i i I i i i Z i ! i i i i : : ~ . - _ _ _ = 6 .......... i . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . t- Y

4 . . . . ............. : . . . . . . . . . . . . . : . . . . . . . . . . . . . i . . . . . . . . . . . i . . . . . . . . . . . . . . . . . . . . . . . .

LL : : i i i : 0 < ~ i ::

lo < ~ . . . . . . ............... ................. i . . . . . . . . . . . . . . . ................ ................

o ~o ICP ELEVATION [era H20]

B LU 20 , F I G U R E 5. Extracted standard deviation of relative spectra us- 1 8 ~ ~ ing the complete data s e t .

'~ 1 t4 ........................................................................................................... mean value within the frequency range. For each spectrum, this measure was observed to follow a roughly linear

1 2 ~ - ~ > . . . . . . . i . . . . . . 1 trend with ICP elevation for different values up to 20 cm w m 10t- ~ ~ . ....... -~ H20 (15 mm H g ) a b o v e normal, where a plateau in the

/ / acoustic response was reached. Thus, a least-squares min- ~;<~ 8 ~ - ~ , , - - - - ....... .-~ imization procedure was used to fit a line passing through

6 k - / ............................................................................................................. -1 the origin to represent the standard deviation v e r s u s ICP 4 ~ - ~ ~ " t . . . . . . "-~ elevation data in each case, and the residuals of this fit

/ ~ ~ / were examined to determine at what ICP elevation the 2 ~ U ~ - ] measure deviated from a linear trend and thus reached a 0~0 10 20 30 40 50 60 70- plateau. In this manner, only standard deviation measures

ICP ELEVATION [el H20] taken before a response plateau was reached in each case were included in the data set pooled across subjects. For

C r~"' 8 this pooled standard deviation data, the correlation coef- 2 ~ } ! ! i ficient and slope of a line passing through the origin were

7 t - ~ . . . . . . . . . . . . . . . . . -'L examined to assess the linearity and variability of the over-

6~ ............................ / ............................................ ;i ~ ........................... -L all response to ICP changes.

................. / ..................................................... :i ................................. t __,s 4 ~ - ~ . ~ ......... -~ A representative set of absolute impedance spectral ee-

l

~ I . . ~ ' ~ 3 " ~ : .................................................................................................... "J t he t ima te s inmagn i tude lZ i@landphase f f 'Z i ( f ) f ~ 1 7 6 measurements sites at normal ICP is shown in

2~- ~ . . . . . . . . . -1 Fig. 2 A, B. The phase change from slightly positive to 7, I I negative in the driving point impedance at very low fre-

l ~ / p ~ ................................................................................................. i I .............. 1 quencies is consistent with the study by Hakansson (4), 0g ]^ L 1̂̂ I ^ / ^ I ^ . . I where the acoustic response of the skull was modeled as a

r u ~v ~,v ~u .~u ov o~, .~ simple resonant structure with mass, resistance, and corn- ICP ELEVATION [el H20] pliance. Also at very low frequencies, all of the responses

have similar phases indicative of nearly in-phase vibra- F I G U R E 4. E x t r a c t e d ( A ) m a x i m a ; (B ) maxima-minima, and (C) standard deviation of the relative spectra of F i g . 3 b e t w e e n 4 tional motion of the entire head. At higher frequencies, the and 7 kHz versus t o t a l I C P elevation using a reduced data s e t . relative motion of the sites becomes out of phase indicat-

ing that the head exhibited a flexural response to the


MAGNITUDE OF RELATIVE IMPEDANCE TO SITE 1

25-- ~ .

20 �9 ~ _ , �9

15"

10-

ICP 5- ELEVATION ~ 1 0 0 0 0 [cmH20] 0 4000 6000 8000

0 2000 FREQUENCY [Hz]

iCP ELEVATION [cmH20]

PHASE OF RELATIVE IMPEDANCE TO SITE 1

10

0

-10

. o

15

10" ~

0 2000 4000 6000 8000 10000

FREQUENCY (Hz I

FIGURE 6. Representative relative transfer impedance (A) magnitude I~(f, ICP)I and (B) phase <~ 7-,4f, ICP) versus-frequency and ICP elevation up to 30 cm H20 (22 mm HG).

broadband excitation. Eliciting a flexural response was a key goal of the experimentation and to our knowledge represents the first experimental measurements o f this phenomenon. This finding lead us to focus our analysis on frequencies well above 1 kHz where flexural modes occur and where we had hypothesized that the response might be most sensitive to changes in ICP.

Figure 3 depicts a representative magnitude of the relative driving point impedance, ]Z(f, ICP) 1, as a function of frequency and ICP elevation in a three-dimensional plotting form. At normal ICP, I Z(f, ICP) I is unity at all frequencies representing the reference spectrum. As ICP increases, a number of spectral changes are observed up to about 20 cm H20 (15 m m Hg) o f elevation where the responses reach a plateau and do not change further. The

magnitude of the maximum spectral changes are roughly 50% of the normal ICP spectral value of unity and are reliably measured given the accuracy of the instrumenta- tion and signal processing. These changes do not occur over time if ICP is not elevated as observed in the control sheep (#3) , where even after 3 h of elapsed time the spectral values were within 4% of unity over the entire frequency range. Also, although the absolute impedance estimates varied with measurements site (as shown in Fig. 2), the relative impedance changes with ICP were consistent and quite similar across sites. This finding indicates that the altered vibrational characteristics due to elevation in ICP were not localized to a region of the head but rather represented a more spatially global response.

In the preliminary data analysis, extraction of such pa-

. . . . . . . . . i#.o t i / . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . .

-~ ~ ::o ~ ?| ? :: i ] :: . . . . . . . . . . . . . . . . . . . . . . . ; . . . . . . . . . . ' . . . . . . . .

~ 2 | : / , ~ , , , ~ . . . . | ;HEEP; II I

ZO .... ! ~ ~ S H E E P ; ......... i ......... i .........

- I I I 2 4 6 8 10 12 14 16 18 20

ICP ELEVATION [cm H20]

~ 5

~3

~2

. . . . CQo~.s . . . . . . . . . . . . . . . . . . . . i . i .

~ ~ ] i ] . . . . . :: . . . . : i ' ~ . . . .

......... i i i, v7 i ! ......... .........

�9 ~ " ~ ~ ~ ! i !! e : | i i i ! i ~ . . | ..... 2 .........

I I l I I 2 4 6 8 10 12 14 16 18 20

ICP ELEVATION [crn H20]

FIGURE 7. Standard deviation of relative (A) magnitude and (B) phase between 4 and 7 kHz versus ICP elevation for the four sheep.


TABLE 1. Sensitivity (slopes) and correlation coefficient data for each sheep and the study group,

Magnitude Phase

Sheep Slope Corr. Coef, Slope Corr. Coef.

4 0.94 0.95 0.46 0.95 5 0.42 0.95 0.30 0,97 6 0.47 0.93 0.30 0.92 8 0.38 0.93 0.23 0.92

Group 0.48 0.92 0.30 0.95

rameters as relative spectral peak height (maxima), peak height relative to valley depth (maxima-minima), and an index of spectral energy (standard deviation) were calculated over the 4- to 7-kHz frequency band using a selected number of measurements (reduced data set) in each case to more efficiently determine trends. Figure 4 A-C depicts the relationships among these three parameters and ICP elevation as extracted from the spectra of Fig. 3. The trend of a roughly linear response of each parameter to ICP elevation up to approximately 20 cm H20 (15 mm Hg) is observed in this case. After extracting these parameters from a few additional sites, the standard deviation parameter was determined exclusively for the remainder of the study as it was less sensitive to the exact choice of frequency range, did not depend upon the exact location of the peak or valley, and was computationally simple. Fig- ure 5 shows the calculated standard deviation values for this case using all of the measurements (complete data set).

Similar spectral changes in proportion to ICP elevation were observed in the phase of the relative impedance,

Z(f, ICP). Figure 6 A and B shows a representative relative magnitude and phase spectra as a function of ICP elevation for the transfer impedance to the # 1 site. Here, the ICP elevation range is limited to 30 cm H20 (22 mm Hg) to highlight the response prior to the plateau. In es- sentially all cases, nearly equivalent information concern- ing ICP change was contained in both the magnitude and phase of the acoustic response from a particular measurement site.

Plots of the standard deviation versus ICP elevation derived from the relative magnitude and phase spectra in each sheep over the 4- to 7-kHz frequency range are presented in Fig. 7 A and B. A linear overall trend with ICP elevation is observed and the correlation coefficients approach unity (>0.92) in each case and for the group. Table 1 summarizes the dynamics of this particular variable for the four sheep in which complete data were obtained. Although the exact range of ICP elevation values over which a linear trend was found varied across sheep, high overall correlations between this acoustic variable and 1CP elevation were observed.

DISCUSSION

The goal of determining if and to what degree the vibrational response of the head changes with elevations in ICP was accomplished in this study. The broadband excitation elicited a flexural response that exhibited changes with ICP elevation up to as much as 20 cm H20 (15 mm Hg); a clinically significant range. Data analysis showed that at least one extractable acoustic parameter, the standard deviation between 4 and 7 kHz, correlated well with ICP change.

The experimental protocol presented a number of tech- nical challenges. Correct catheter placement near the center of the lateral ventricle was essential to insure catheter patency. The attachment of pumps for infusion at very slow rates immediately after placements also acted to keep the catheters patent. However, the possibility remains that this procedure caused some increase in ICP above normal to occur before the acoustic measurements began. This may explain why the acoustic response reached a plateau earlier in some of the sheep, as their ICP may have been prematurely elevated to some degree. This phenomenon was evident in sheep #7 where a surgical error caused a hemorrhage that increased the baseline ICP to almost 45 cm H20 (33 mm Hg). Changes in the acoustic response of this animal due to small ICP elevations above this high baseline were relatively diminished, as was observed in the response plateau of the other sheep. Thus, the data from this sheep were not included in any subsequent analysis.

The use of a specially constructed shaker mount and frame was necessary to insure that the weight of the shaker did not significantly load the skull, and that there was no bulk movement of the head. Also, tapping a hole in the skull and using epoxy for each transducer attachment pro- vided direct measurements of the skull's vibration without the potentially confounding effects of the extracranial tissues.

The observed responses were a strong function of frequency and a much weaker function of the measurement site. The analytical focus was placed on the 4- to 7-kHz frequency range primarily because we had hypothesized that flexural modes of the head might be most sensitive to ICP changes. Initially, parameters such as peak height were extracted and related to ICP change, but the standard deviation measure proved most robust since it was less sensitive to the exact relative spectral shape changes with ICP elevation, and was also computationally simple to determine. As is evident by the wealth of spectral changes that occurred, there may be other extractable measures that will prove to be even more reliable and sensitive indices of elevated ICP.

Relating the measurable response of the head directly to underlying structural changes due to ICP elevation is dif-


ficult for frequencies above 1,000 Hz. However, the general trend of increasing impedance magnitude with ICP suggests that the overall stiffness of the head increased in proportion. The response plateau cannot be explained sim- ply though, and this observation suggests that small pressure increases above normal resulted in some net expan- sion or relative movement of the skull and brain which stops after reaching an elastic limit. More detailed analy- ses of the measured response, especially at lower frequencies where more direct ties to structure can be made, are needed to fully understand and interpret these positive overall experimental findings. In addition, broadening the study to investigate the ability to monitor not only increases but also decreases in ICP in an intact head is required before the full clinical utility of this approach can be determined.

REFERENCES

1. Bekesy, G. V. The mechanical properties of the ear. In: Handbook of Experimental Psychology, edited by S. S. Stevens. New York: John Wiley, 1951, pp. 1075-1114.

2. Franke, E. K. Response of the human skull to mechanical vibrations. J. Acoust. Soc. Am. 28:1277-1284, 1956.

3. Gurdjian, E. S., V. R. Hodgson, and L. M. Thomas. Stud- ies on mechanical impedance of the human skull: preliminary report. J. Biomechanics. 3:239-247, 1970.

4. Hakansson, B., and P. Carlsson. The mechanical point im-

pedance of the human head, with and without skin penetra- tion. J. Acoust. Soc. Am. 80:1065-1075, 1986.

5. Hamilton, A. J., P. M. Black, J. Madsen, and D. B. Carr. Neuropeptide investigations with an ovine surgical model. Neurosurge~ 18:748-755, 1986.

6. Hamilton, A. J., A. Cymmerman, and P. M. Black. High altitude cerebral edema. Neurosurgery 19:841-849, 1986.

7. Hodgson, V. R., E. S. Gurjian, and L. M. Thomas. The determination of the response characteristics of the head with emphasis on mechanical impedance techniques. Proc. llth Stapp Car Crash Conf., 1967, pp. 125-138.

8. Hodgson, V. R., and L. M. Patrick. Dynamic response of the human cadaver head compared to a simple mathematical model. Proc. 12th Stapp Car Crash Conf., 1968, pp. 280- 301.

9. Kay, S. M. Modern Spectral Estimation: Theory and Ap- plication. New York: Prentice-Hall, 1988, pp. 76-77.

10. Khalil, T. B., D. C. Viano, and D. L. Smith. Experimental analysis of the vibrational characteristics of the human skull. J. Sound Vibrat. 63:351-376, 1979.

11. Luchka, S. Working with ICP monitors. R.N. April: 34-37, 1991.

12. Pappenheimer, J. R., S. R. Heisey, E. F. Jordan, and J. DeC. Downer. Perfusion of the cerebral ventricular system in unanesthetized goats. Am. J. Physiol. 203:763-774, 1962.

13. Semmlow, J. L., and R. Fisher. A noninvasive approach to intracranial pressure monitoring. J. Clin. Eng. 7:73-78, 1982.

14. Stalnaker, R. L., J, L. Fogle, and J. H. McElhaney. Driv- ing point impedance characteristics of the head. J. Biomech. 4:127-139, 1971.

the effect of elevated intracranial pressure on the vibrational response of the ovine head

Documents