A computerized kinematic diagnostic system

Journal of Medical Systems, Vol. 13, No. 5, 1989

A Computerized Kinematic Diagnostic System Richard H. Eckhouse, Jr., Ph.D.*, Ruth A. Maulucci, PhoD.t, and Elizabeth Leonard, Ph.D.~

This paper describes the instrumentation developed for a clinical system that measures upper extremity kinematics in normal and cerebral palsied children. The ability to diagnose cerebral palsy during early and midinfancy is influenced by the fact that movement and postural abnormalities become apparent only over time, and are not readily detectable until there is sufficient abnormality so that it can be viewed by gross inspection during clinical examination. The methodology presented here serves to discriminate normal from deviant movement at an earlier age than what is presently possible, and may also result in alternative therapeutic intervention.

INTRODUCTION We have developed a clinically based system that measures upper extremity kinematics in normal and cerebral palsied children, and demonstrated its clinical feasibility and utility. Such a system is valuable in that it permits greater understanding of the development of movement dysfunction in cerebral palsy. Development of perinatal and neonatal medicine has progressed so that extremely premature infants are surviving. These infants have a high probability of developing intraventricular and intracerebral hemorrhages which in turn produce disorders in postural control and movement. Early identification and treatment of infants with established disability or those at risk is now mandated by Public Law 99-457 and present methods for early detection and treatment of cerebral palsy remain largely untested. The ability to diagnose cerebral palsy during early and midinfancy is influenced by the fact that movement and postural abnormalities become apparent over time and that this condition is not readily detectable until there is sufficient abnormality that can be viewed by gross inspection during clinical examination. The utilization of magnetic resonance imaging and cranial ultrasonography during the neonatal period and early infancy can alert the physician to the presence of periventricular leukomalacia and intraparenchymal hemorrhages that lead to suspicion that development of cerebral palsy is a high probability. From the *Department of Mathematics and Computer Science, UMASS-Boston 02215. ~'MOCO, Inc., Scituate, Massachusetts. ~:Pediatric Rehabilitation Research Laboratory, Children's Health Center, St. Joseph's Hospital and Medical Center, Phoenix, Arizona. 261 0148-5598/89/1000-0261506.00/0 © 1989 Plenum PublishingCorporation

262

Eckhouse et al.

However, the diagnosis of cerebral palsy is generally not made until clinical signs of motor dysfunction are present as revealed by persistence of primitive reflexes, delayed motor milestones, or development of disorders in muscle tone, posture, and coordination. Our kinematic diagnostic system enables precise quantification of movement that is not possible with the pediatric neurological examination or existing motor development scales which are, at best, gross measures of motor function. Kinematic analysis detects subtle motor dysfunction and permits precise quantification of temporal and spatial parameters that underlie movements. Our methodology is intended to discriminate normal from deviant movement at an earlier age than what is presently possible, and may result in alternative therapeutic intervention, hence contributing to an improved understanding of the ontogeny of cerebral palsy.

BACKGROUND

An expanding corpus of research exists presently on development of infants subjected to biological stresses resulting from prematurity, asphyxia, intrauterine growth retardation, central nervous system hemorrhage, infection, and genetic disorders. Problems associated with early identification, diagnosis, and therapeutic management of developmentally vulnerable infants are complex. In regard to motor handicaps, relatively little experimental research has investigated the early identification and treatment of infants with cerebral palsy and related motor disorders under controlled experimental conditions. 1-3 The number of infants surviving prematurity and catastrophic neonatal illness has focused increased attention on the long-term sequelae of neonatal outcome, particularly impaired developmental function. Over the past 15 years dramatic improvements in neonatal mortality and morbidity have occurred. 4-7 Recent studies, however, reported an alarming incidence of neurologic morbidity for very low birth weight survivors and raised significant questions regarding the quality of life of these babies. Schechner8 indicated that 44% of infants with birthweights less than 1000 grams had developmental disabilities including motor handicaps characterized by spastic quadraplegia, hemiplegia, or diplegia. Knobloch e t a l . 9 reported that 40% of infants with birthweights between 751 and 1000 grams had motor disorders characterized by spastic quadraplegia and mixed cerebral palsies. Vohr and Hack, 7 in an extensive review of developmental follow-up of low birthweight infants, showed that the incidence of neurological morbidity for infants between 1000 and 1500 grams ranged between 13 and 19%; the incidence of neurologic deficits climbed to 49-55% for survivors less than 800 grams. Demographic data show that the overall incidence of cerebral palsy among infants with birthweights over 1500 grams has remained relatively stable. 7'9 A decrease in cerebral palsy among infants with birthweights over 1500 grams has occurred." This has been counterbalanced, however, by the high incidence of neurologic deficits in surviving very low birth weight infants. Thus, the number of infants surviving pre- and perinatal complications who develop motor handicaps are cause for concern. Cerebral palsy, a motor disability originating prenatally and during infancy, pfoduces disorganization of motor function resulting in spasticity and/or involuntary limb movement. ~o Cerebral palsy is caused by impairment of brain structures subserving pos-

A Computerized Kinematic Diagnostic System

263

ture and movement and occurs in 0.06-5.0 out of 1000 births. ~1 The condition is nonprogressive, but the degree of motor disability can be severe. Relatively little is known about the ontogeny of cerebral palsy. Physical findings include, in part, alterations in muscle tone, spasticity, athetosis, and disorders of posture, equilibrium, coordination, and skilled movement. Because a diagnosis of cerebral palsy connotes life-long impairment, it is important to be able to identify successfully infants who are likely to have permanent impairment as early as possible. Procedures in current clinical use are not presently able to accomplish this with a high degree of reliability prior to twelve to eighteen months of age. Cerebral palsy is suspected when at risk infants fail to achieve anticipated motor milestones, notably, independent sitting between 6 and 8 months and independent ambulation between 9 and 15 months. Delays in motor development may be accompanied by alterations in muscle tone and postural deviations. The assessment of tonal and postural abnormalities in infants is highly subjective and not clinically reliable thereby reducing the ability to establish, with certainty, a diagnosis of cerebral palsy prior to one year of age, except in the most severe cases. Prior to 12 to 18 months, two important factors impact on the ability to diagnose cerebral palsy. First, some infants may show transient neuromotor dysfunction that resolves spontaneously. Second, physicians may be uncomfortable attaching a diagnostic label to neuromotor irregularities because the clinical signs of increased tone, hyperactive tendon reflexes, and postural asymmetry may be suggestive of dysfunction but not necessarily diagnostic.

Infant Evaluation Scales Little research has been directed toward investigating the validity of child development scales for diagnosing motor dysfunction. Motor development scales such as the Bayley Motor Scale, 12 the Denver Developmental Screening Test, 13 and the Gesell Developmental Schedules 14 document the cephalocaudal progression of motor skills and are hierarchically organized. The standard neuropediatric examination identifies neurologic abnormalities 15'~6 through systematic evaluation of the central and peripheral nervous systems but does not have well established predictive validity for infant motor disorders. 17 In infancy, a medical diagnosis of motor abnormality is based largely on the retention of primitive reflexes, tonal abnormalities, and failure to meet expected motor milestones. The current practice for evaluating neurological development provides for recurrent assessments during the first few months of life with the intent that frequent evaluations will detect changes in motor behavior from one examination to another. Tests currently used for developmental assessment present a potpourri of items that measure different behaviors that include but are not restricted to motor skills. In these tests, motor performance items are widely spaced and may not be sufficiently discriminative for detection of motor abnormalities. Few scales measure adequately qualitative differences in motor performance. For example, there is presently no reliable way to measure how spasticity or athetosis influence movement execution. Traditional measures quantify motor behavior by recording the age at which a child passes selected motor milestones. For example, the ability to stack two cubes, an item passed at a mean age of 13.2 months according to Bayley 12 norms, chronicles the age at which an infant performs this behavior. Qualitative aspects of this motor behavior, such as hand orientation or the

264

Eckhouse et al.

presence of a mild tremor, are ignored by the scoring criteria. If an infant can perform this task he/she passes, if not he/she fails. An infant with spastic quadraplegia may be able to perform this task but may do so with a flexed wrist and pronated forearm, an abnormal reaching pattern. That infant will, nonetheless, receive credit for the task. The test is insensitive in discriminating how individual test items are executed. It is possible, therefore, to have test scores that are artificially inflated that do not reflect accurately whether motor dysfunction may be present. Reliable identification of impaired motor function in infants and children is acknowledged to be difficult. In fact, the ability of the pediatric neurological examination to quantify motor function was questioned, is The pediatric neurological examination is confronted with serious difficulties in determining what constitutes normal responses. Criteria for determining normality in responses such as muscle tone and movement are poorly defined and may vary over time and with examiners. As a result, conventional medical approaches for diagnosing motor dysfunction during infancy and early childhood are not specific enough to quantify motor abnormalities in sufficient detail. Given these limitations, it would appear that a quantifiable technique for evaluating motor function could offer a specific clinical advantage, namely a high degree of test and clinical reliability. Kinematics

Kinematics deals with mathematical methods of describing the temporal and spatial aspects of motion. Kinematic procedures can be used to quantify several human performance parameters such as the position, velocity, and acceleration of various limb segments and joint angles. For example, kinematic analysis of upper extremity function yields the trajectories of the limb components as the hand traverses three-dimensional space. Quantifiable data on the temporal and spatial characteristics of these trajectories, such as movement time, accuracy, maximum and minimum values, and the time of occurrence of the extrema, can then be extracted. In contrast to the pediatric neurological examination and traditional child development scales used to measure motor function, kinematic procedures permit more precise, fine-tuned analysis of movement, are highly reliable, and are not subject to examiner interpretation. Furthermore, kinematic analysis is able to detect subtleties of movement that cannot be ascertained by visual inspection. For example, during a physical examination a physician may subjectively judge how well a child moves his arm when reaching for an object; however, only gross movement, at best, can be evaluated in this manner. Precise quantification of the defining parameters of the movement, such as peak velocities and accelerations, can only be acquired by quantitative recording instruments. The use of kinematics to study pediatric motor disorders has not been investigated to our knowledge. Groundbreaking work by Forssberg, 19'2° Thelan, 21'22 and von Hofsten23-25 employed kinematic analyses to study motor development in normal infants. The technologies used in these studies have some significant limitations that present severe restrictions if these procedures are to be adapted for diagnostic purposes for disabled populations. The two principal constraints are the length of processing time required to analyze biological signals and the number of samples required to generate meaningful data.

A

Computerized Kinematic Diagnostic System

265

Videotape was used by Thelan to record leg movement in normal infants. She later analyzed these recordings off-line using frame-by-frame analysis. This technique was used to investigate the typography of infant kicking but required extracting data from literally thousands of frames, thus limiting its potential usefulness as a diagnostic procedure. Von Hofsten studied the development of reaching and grasping in infants ages 12-18 weeks. Limb movement was transcribed from videotape to a computer with an X-Y plotter that translated movement into horizontal and vertical coordinates. Data were generated on the limb trajectory including hand velocity, acceleration, and the number of movement sequences used to contact a target. The processing time required to reduce raw data into useful kinematic profiles was exceedingly long. Forssberg studied reflexive stepping and the transition to bipedal locomotion in normal human infants during treadmill locomotion. He employed the Selspot system (Selcom, Partille, Sweden), an optoelectronic device utilizing light emitting diodes (LED's), to record lower extremity limb position in the saggital plane. Herman and Leonard26 studied lower extremity rhythmical behavior using the Selspot to record supine kicking and independent locomotion in infants ages 4--12 months. Experimental studies currently being performed by Helxnan, Maulucci, Leonard, and Psyzka27'28 using the Selspot system are able to record limb movement in adults during functional activities such as kicking, locomotion, and reaching to a target. Additional studies by Maulucci, Jerard, and Herman used the Selspot to analyze arm trajectories in normal adults. Set-up time and processing time with the Selspot system is prohibitively long for clinical diagnostic use.

SYSTEM AND INSTRUMENTATION The instrumentation for this research involved constructing a laboratory prototype of the kinematic diagnostic system (KIDS), consisting of a specially designed positioning chair; an automated device for target presentation; kinematic sensors, called ISOTRAK's, for data acquisition; and the linking of all of these components to a PC-compatible microcomputer. Software was developed to interface the devices to the microcomputer for experiment control and data acquisition.

The Positioning Module The positioning module is a specially designed chair that positions the head, trunk, and arm optimally to execute the desired limb movement. This module has four degrees of freedom to facilitate optimal positioning and counteract the deforming forces produced by spas~icity. This chair allows stabilizing head position in both roll and pitch, i.e., in the anterior2postefior and lateral planes. The angle of the torso segment is also adjustable to control trunk hyperextension. The hip angle is adjustable to counteract extensor spasticity and thrusting of the lower extremities. The entire positioning module is reclinable. Finally, the design involved an important consideration, namely the chair could contain no metal or it would interfere with the ISOTRAK sensing. For the purposes of recording the time of initiation of arm movement, a copper

266

Eckhouse et al.

touchplate was attached to the armrest of the chair. Arm contact with the touchplate was required before a trial would begin in order to assure a standard starting position, and such contact is sensed by the microcomputer. The Terminal Reaching Device An electronic jack-in-the-box was constructed to present the reaching target. Several different mechanical as well as electronic designs were tried and rejected, because they could not be controlled with sufficient resolution or were too distracting or simply did not work. The final solution was quite simple. A ball with a 5-cm radius was attached by Velcro to a V2-in. wooden dowel that connected to a cylinder. The ball was wrapped in touch-sensitive material, so that a touch of the ball could be detected by the computer and the time of the touch recorded. This apparatus was housed in a plywood casing and was set into a wooden base so that it could be moved easily into position. The unit was painted flat black to be unobtrusive. The cylinder was connected to a wall unit for compressed air which was used to force the rise and fall of the ball. A minimum pressure of 5 psi was required for activation. The electronics for controlling this device represents an inexpensive solution. A circuit diagram for the electronics is shown in Fig. 1. The entire jack-in-the-box was moveable so that it could be independently located to permit measuring both in-plane and out-of-plane movement. The final position of the exposed object, the ball, was configured so that the child was required to orient the hand in a specific direction to maximize contact and minimize trajectory variability. The Kinematic Recorder After reviewing and working with several devices that are capable of monitoring and recording spatial coordinates, we selected a system called the 3SPACE ISOTRAK manufactured by Polhemus Navigation Sciences. 29 This inexpensive unit is easy to use, and represents a great simplification over existing devices. The most commonly used recording device for acquiring kinematic data is the Selspot system manufactured by Selcom, Partille, Sweden. Experimental paradigms have been developed by both Leonard and Maulucci 28 to use the Selspot system to acquire normative data that could be compared with data derived from subjects with neurological disorders affecting movement. The Selspot is a sophisticated precision system, ideally suited to sensitive laboratory experiments. It is not appropriate, however, for clinical diagnostic purposes or for young subjects for the following reasons. The LED's are glued to the skin and connected with long wires to a processing unit, making the procedure uncomfortable and cumbersome for children. The Selspot requires lengthy calibrations and processing time to convert raw data into useful kinematic profiles, as well as requiring a trained technician to operate it; these factors make it unsuitable for a clinical environment. The development of the 3SPACE ISOTRAK system offers some marked improvements over optoelectronic and videotape methods for quantifying movement behavior. The ISOTRAK unit is a real-time, electromagnetic feedback mechanism that provides the position and orientation, i.e., 6 degrees of freedom (df) of a moving sensor relative to a fixed source. The technology uses low frequency (2-14 Hz) magnetic fields to make these

A

Computerized Kinematic Diagnostic System

8

I

7

I

6

o,o

,~-~

i

o_

F~)

s

i

~

*

!

~

'ZeM

~

~

~ OZ~ 4e13 ~

f---ff

~

s".~~

~%. Nl~ck

i

_.~RS 2~

1

~

~

8i

.....

!CZ

l

C~ble coloez~ 1. ~ 1 ~ Z- Red 3. Bl~ck ~. Green 5. 81ue

~.lu~ Q

"•e.lu$ -~

~

i:

~

P~:~--'-'--~-

z

~.~.

tiC~ 3- ~,,, I ~I

:R2

i

Jl

1°~

~.

~~.~

~1~

Red

2~7

"

Un,__.u• e d 1~a~o~ # nZC 84eel

R

[

~

s

I

7

I

6

~ ~

I

s

I

~

~ [ o~, I

MOCO.,nc.

~.o. ~ x ~ ~ . ~ ~z~s TP/PLUNGER CIRCUITS

~ , ~ ~z~. 3 ~

~

~ ~~ ~ [ ~

~~

I ~

FiSur¢ L Touch circuit e]ec~on~cs and #neumatic cyfi~dCr c o ~ o l .

measurements. Output is in the form of time series signals which uniquely determine the position and orientation of the sensor relative to the source. The sensor is a small 0.59-ino cube and weighs 0.5 oz. The source assembly measures 1.8 in. high by 1.7 in. square and weighs 8.6 oz. The system provides hemispherical measurement within a radius of 32 in., and angular accuracies in excess of 0.5 ° at the 50% confidence interval throughout a compensated operational envelope. System resolution is approximately 0.1 ° angular and 0.05 in. translational. Sampling rate is 60 Hz per unit, i.e., 60 samples per second total for a single sensor and source and 30 Hz when two sensors are used. The output is connected via an RS232C port to a microcomputer for processing. This electromagnetic measurement technique can be used as a feedback mechanism in closed looped operation and provides simple, precise 6 df measurement at a rate sufficient for use in a diagnostic and training environment. Each ISOTRAK sensor allowed the recording of the three rectilinear (x, y, and z) and the three angular (azimuth, elevation, and roll) coordinates of the rigid body, i.e., the arm

268

Eckhouse et ai.

segment, to which it was attached. Thus, the exact location and orientation of the ann segment, and ultimately of the entire ann, was specified throughout the interval of motion. The Microcomputer

The touchplate, ball, pneumatic cylinder, and two ISOTRAK's were interfaced to a Compaq Deskpro microcomputer for real-time control and data acquisition. Temporal resolution for the touchplate, ball, and cylinder was 50 msec. Sampling rate for the ISOTRAK's was 28 Hz each. This is generally considered somewhat slow for human movement studies but worked surprisingly well (see Results). Software was developed for simultaneous data acquisition from the two ISOTRAK's. Although we had implemented such software for a minicomputer, it has not previously been accomplished by anyone for a microcomputer. The software is a combination of assembly language and C code, and includes computer-control of the target presentation from appearance to disappearance, as well as data acquisition. Data acquisition consisted of the six trajectories from each ISOTRAK, from the appearance of the ball to the touch of the target. Two temporal parameters were also recorded via the touchplate and the touch sensitive ball, namely, reaction time and movement time. Reaction time is the interval between the time of presentation of the target and the time at which the child's arm leaves the touchplate. Movement time is the interval between the lifting of the arm from the touchplate and the time of contact with the ball. All data were stored on a hard disk for later analysis.

RESULTS Research Protocol

A study was conducted to determine whether the KIDS can discriminate in a group of three year olds between children with diagnosed cerebral palsy and normal children by means of their arm trajectories. We intentionally selected this age group in which cerebral palsy has already been diagnosed by conventional methods, in order to determine under established conditions what the discriminating features are between the known normal and abnormal trajectories that will be diagnostic of pathology. We are currently in the process of repeating this study with progressively younger age groups, ultimately determining whether the system is capable of extracting sufficient information from arm trajectories to predict the occurrence of cerebral palsy in high risk children prior to conventional medical diagnosis. Fifteen tfitee year old children participated in the clinical trials. Eight normal children and seven children with spastic diplegia were tested. Each child was seated comfortably in the positioning chair. Initially, we planned to restrain the shoulder so that the biomechanical model would be simpler, with seven degrees of freedom. However, the children became quite agitated by this restraint, and were not making natural arm movement. Therefore, we allowed the shoulder to make translational movements, producing a ten degree of freedom model.

A ComputerizedKinematic Diagnostic System

269

One ISOTRAK sensor was placed on the dorsum of the hand just inferior to the third metacarpal phalangeal joint, and another on the upper arm at a point midway between the antecubital fossa and the acromium. With this placement of the sensors, we were able to monitor all of the ten independent coordinates plus two redundant ones, so that all ten degrees of freedom of the system were specified. The cylinder shaft containing the target, i.e., a silver sphere, was positioned in front of the child, either in plane with the shoulder or at midline. Prior to each trial, the child was instructed to rest his arm on the touchplate attached to the chair, watch for the ball, and reach to touch it. Once the child directed his attention to the shaft, the computer operator initiated the trial with a keystroke. Each trial began with the appearance of the ball and terminated when the child touched the ball, causing it to disappear again. If a child could not make contact with the ball, the trial automatically terminated after 30 sec. All events and data acquisition were computer-controlled. Sixteen trials were presented to each arm, eight trials in-plane and eight at midline, for a total of 32 trials for each child. The twelve trajectories and two temporal parameters were collected and stored for each trial. The total testing time including subject preparation and recording took approximately 30-40 rain.

Analysis The biomechanical model is a 10 df model, with the degrees of freedom being the three translational coordinates of the shoulder, shoulder abduction and flexion, humeral rotation, elbow flexion, radial rotation, wrist flexion, and radial-ulnar deviation. With the sensors placed as described earlier, it is possible to select a subset of ten independent coordinates from the set of twelve. For example the six coordinates of the upper arm sensor, the three rotational coordinates of the hand sensor, and one of the translational coordinates of the hand sensor completely specify the system, with the forearm and the remaining hand coordinates being uniquely determined from these specified coordinates. The twelve ISOTRAK trajectories represent a system that is overdetermined, in that two of the trajectories are unnecessary. However, these were used as a measurement of the errors introduced by the hardware, as well as by various calculations and assumptions. Note that it is not necessary to use functional angles, such as elbow flexion, in the model, since any set of independent coordinates will serve the same purpose as any other set in terms of discriminating arm movement. It would be interesting to view these functional angles and it is possible to calculate them from the existing trajectories. This will be accomplished in future studies. The data were analyzed by means of the ASYST Version 3.0 software package. While this is cumbersome to use and requires a nonintuitive thought process, it is nonetheless a comprehensive package that incorporates graphics, signal processing , and statistics to a degree that no other package that we are aware of does. For each trial, the twelve trajectories were plotted in raw form, except that for each trajectory, the average value of the arm position before it began to move was calculated and used as a zero offset so that intra- and intersubject trials could be compared. In the interest of economy, and also to facilitate visual inspection, these trajectories were plotted three to a graph, with the rectilinear coordinates overlaid on a single plot for each of the two sensors, and similarly for each sensor the angular coordinates overlaid on a single plot.

270

Eckhouse et al.

~ ×E

-[:~104. N19

A

~

il.2

8.45

5,73

3. ~e

.~-73 -

xE e

U

_

-LI ~ 4 . NJ.9

A

~~'~'"~!~ ~,~.~ .....:~.........~~.......... ~....... ~~::~~~ ........... ~.~...~~ .......... . .~.~. .,:............. ~~ ~~.......... ......... . ~:.......... ......::~..........~~:: -....+: -""

:

.........

X ........ ~.................. I

~

4.'1~

~

~

:

~................... ~......... ~......... ~.......... ~.......... ~.......... ~ ~

12'.3

~

'

-

'

~2~'. S

~

28~.?

~

~61.9 ~E

~

F i ~ e l. Si~n~l~ ~ o m a t y p i ~ i n - p ] ~ ~ 1 ~f ~ n ~ l ~ub]ec~. (~) r e ~ t i l i n ~ ~o~din~te~ of the h ~ d . ( ~ ~ u l ~ ~ o ~ r d ~ t ~ ~ ~h~ h ~ d . (C~ rectiline~ ~o~din~te~ ~f ~ e upper ~ . (D~ ~ g ~ ] ~ coordin~[~s ~ thc up~ ~.

A Computerized Kinematic Diagnostic System

C xE

271

~I~4.N19 B

~ 2,65

1.52

.397

-.729

-1.85

D

~I~4.N19

xE

~

xE

~

B

~:E ~ II.5

5. ~6

-i. 41

-?,88

-14.3

Figure 2. Continued.

Several general signal characteristics were observed with regard to the shape of the individual trajectories. This was true not only across trials from a single subject but also between subjects and in some cases even across the two groups. For example, the concavity and inflection points of the rectilinear coordinates are very similar across all

272

Eckhouse et ai.

subjects. Furthermore, certain of the signals, such as the azimuth and roll of the hand, are phase-locked. In contrast, other characteristics showed marked differences between the normal and cerebral palsy groups. For example, the angular coordinates for both the hand and the upper arm display far more activity in the cerebral palsy children than in the normals, indicating a more pronounced rotational component in the reach of the former group. Also, although there are several pairs of signals that appear to be phase-locked for all subjects, the actual form of the trajectories of these pairs differs between the two groups. The four graphs of the twelve signals from a typical in-plane trial of a normal subject are illustrated in Fig. 2. Figures 2A, B, C, and D are the rectilinear coordinates of the hand, the angular coordinates of the hand, the rectilinear coordinates of the upper arm, and the angular coordinates of the upper arm, respectively. The ordinates represent inches or degrees, and the abscissas represent time, with 1 unit of measurement being equal to 1/28 sec. For each of the twelve signals in each trial, the displacement, the path length, and the number of reversals of direction were calculated. Each signal was then differentiated to obtain the corresponding velocity signal, and the maximum velocity, the percentage of the interval at which the maximum occurred, and the average velocity were calculated. Additionally, the reaction time and movement time were tabulated. Descriptive statistics were calculated for each of the defined parameters. An analysis of variance was also done for each of these variables, using a 2×2 factorial design with diagnosis and plane. Preliminary results corroborated the informal observations made from the graphics, and also revealed a significant difference in the temporal parameters of reaction time and movement time between the two groups. Moreover, the two groups showed marked statistical differences in virtually all of the kinematic parameters studied.

DISCUSSION Current neurological and developmental evaluation procedures lack sensitivity and predictability for diagnosing pediatric motor disorders. Using the KIDS to examine differences in motor performance between normal and motor impaired populations permits more sensitive analysis of movement parameters, better identifies problems of disordered movement function, and may lead to the development of new diagnostic and treatment procedures. The results of our initial study are encouraging. The feasibility of the hardware and software technology developed during the project was unquestionably established. All solutions are inexpensive and accessible to the individual clinician. The system is easy to use as evidenced by the rapidity with which the therapists learned to operate it. Both parents and children were quite positive about the system and the procedures. In fact, a video tape study showed that the system was easily accepted by the target population as seen from the compliance and enjoyment exhibited by the three year old children with cerebral palsy that were used in the protocol. We are presently expanding this instrumentation and biomechanical procedures to discriminate normal and cerebral palsy children prior to diagnosis by conventional measures. We are also examining the potential of the system as a treatment device. Because

A ComputerizedKinematic Diagnostic System

273

it monitors movement in real time, the system can administer immediate feedback pertaining to the trajectory, such as employing visual or auditory means to provide knowledge of results or knowledge of performance, which can be used to induce trajectory modification. Once this methodology has been refined, applications for diagnosing other forms of motor impairment can also be explored. For instance, kinematic analysis of limb movement will permit evaluation of muscle strength during functional activity in disorders where weakness is a primary clinical symptom, such as the muscular dystrophies. When these diagnostic procedures are developed and clinically validated, their utility for evaluating motor function in other age and disability groups can be further investigated. It is entirely conceivable that this instrumentation and approach to the detection of motor abnormalities will generalize to a wide variety of clinical applications to include diagnosis of neurological and muscular disorders and evaluation of performance capabilities of the upper limb during a wide variety of functional tasks. This includes the use of the KIDS for evaluating functional impairments of the upper limb in adults resulting from orthopedic, neurological, and neuromuscular conditions produced by disease or injury.

ACKNOWLEDGMENT This research was supported by the National Institutes of Health, National Institute of Child Health and Human Development, under Grant 1 R43 HD22660-01A1.

REFERENCES 1. Piper, M., Kunos, V., Willis, D., Mazer, B., Ramsay, M., and Silver, K., Early physical therapy effects on the high risk infant: Randomized control trial. Pediatrics 78, 2, 216-224, 1986. 2. Ferry, P., On growing new neurons: Are early intervention programs effective. Pediatrics 67, 38-41, !981. 3. Committee on Children with Disabilities, School age children with motor disabilities. Pediatrics 76, 4, 648-649, 1985. 4. Nelson, K., and Ellenberg, J., Children who outgrew cerebral palsy. Pediatrics 69, 529-536, 1982. 5. Ellenberg, J., and Nelson, K., Early recognition of infants at high risk for cerebral palsy: Examination at age four months. Devel. Med. Child Neurol. 23,705-716, 1981. 6. Vohr, B., Follow up of neonatal intensive care graduates. Unpublished manuscript, 1983. 7. Vohr, B., and Hack, M., Developmental follow up of low birthweight infants. Pediatr. Clin. North America 29, 1441-1455, 1982. 8. Schechner, S., For the 1980's: How small is too small? Clin. Perinatol. 7, 135-144, 1980. 9. Knobloch, H., Malone, A., Ellison, P., Stevens, F., and Zdeb, M., Considerations in evaluating changes in outcome for infants weighing less than 1501 grams. Pediatrics 69, 285-296, 1982. 10. Swalman, K.F., Cerebral palsy: Perinatal motor impairment. In (K.F. Swalman and F.S. Wright, eds.), The Practice of Pediatric Neurology, Mosby, St. Louis, pp. ~.n.~ 452. ll. Stanitsky, C.L., Cerebral palsy. In (A. Ferguson, ed.), Orthopedic Surgery in Infancy and Childhood, Williams and Wilkins, Baltimore; pp. 665-707. 12. Bayley, N., Manual for the Bayley Scales of Infant Development. New York: The Psychological Corporation, 1969. 13. Frankenberg, W., and Dodds, J., The Denver Developmental Screening Test, University of Colorado Medical Center, 1968. 14. Knobloch, H., Malone, A., and Stevens, F., Manual of Developmental Diagnosis: The Administration and

274

15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.

28. 29.

Eckhouse et al.

Interpretation of the Revised Gesell and Amatruda Developmental and Neurological Examination, Harper and Row, Haggerstown, 1980. Paine, R.S., and Oppe, T.E., Neurologic examination of children. Clinics in Developmental Medicine, William Heinemann, Ltd., London, 1966, pp. 20-21. Prechtl, H., The neurological examination of the full term newborn infant. Clinics in Developmental Medicine, Lippincott, Philadelphia, p. 63. Self, P., and Horowitz, F., The behavorial assessment of the neonate: An overview. In (J. Osofsky, ed.), Handbook of Infant Development, Wiley, New York, pp. 126-164. Taft, L., and Barabas, G., Infants with delayed motor performance. Pediatr. Clin. North America 29, 137-149, 1982. Forssberg, H., and Wallberg, H., Infant locomotion: A preliminary movement and electromyograhic study. In (K. Berg and B. Erikson, eds.), Children and Exercise IX, University Park Press, Baltimore, 1980. Forssberg, H., Ontogeny of human locomotor control I. Infant stepping, supported locomotion, and transition to independent locomotion. Exp. Brain Res. 57,480-493, 1985. Thelan, E., Bradshaw, G., and Ward, J., Spontaneous kicking in month-old infants: Manifestation of a human central locomotor program. Behav. Neural Biol. 32, 45-53, 1981. Thelan, E., Developmental origins of motor coordination: Leg movements in human infants. Devel. Psychobiol. 18, 1-22, 1984. Hofsten, C. Von., Development of visually directed reaching: The approach phase. J. Hum. Move. Studies 5, 160-178, 1979. Hofsten, C. Von., Predictive reaching for moving objects by human infants. J. Exp. Child Psychol. 30, 369-382, 1980. Hofsten, C. Von., Foundations for perceptual development. In (L. Lipsett, ed.), Advances in Infancy Research, Vol. H, Albex, New Jersey, 1983. Herman, R., and Leonard, E., Unpublished data. Herman, R., Maulucci, R., Leonard, E., and Pyszka, V., The initiation of locomotion in man. In (S. Grillner, P.S.G. Stein, H. Forsberg, D. Stuart, and R. Herman, eds.), Neurobiology of Vertebrate Locomotion, McMillan, London, 1986. Leonard, E., and Maulucci, R., Unpublished data. Polhemus Navigation Sciences Division, McDonnell Douglas Electronics Company, 3SPACE ISOTRAK Users Manual, 1985.