Objective. The electroencephalogram (EEG) is widely used in full-term infants with acute neonatal encephalopathy, and its prognostic value has been confirmed by several studies. Magnetic resonance imaging (MRI) of the brain has also been applied in these patients, and increasing numbers of reports affirm its prognostic reliability. The aim of this study has been to investigate the correlation between an early EEG and MRI findings in infants with acute neonatal encephalopathy and to assess the prognostic value of a combination of EEG and MRI findings.
Participants and Methods. Twenty-five full-term infants had an EEG recorded within the first 72 hours after birth and a neonatal brain MRI scan after the end of the first week.
Results. Both EEG and MRI were predictive of outcome. A normal MRI was always associated with normal EEG background activity and normal outcome and severe abnormalities on MRI with marked EEG abnormalities and an abnormal outcome. When the MRI showed moderate abnormalities, the EEG in all cases but one identified patients with normal and abnormal outcome.EEG, MRI, HIE, neurodevelopment.
Hypoxic-ischemic insults are the most common cause of brain lesions in the full-term newborn infant.1–2 Brain imaging, especially magnetic resonance imaging (MRI), has allowed the identification of various patterns of brain lesions in infants with hypoxic-ischemic encephalopathy (HIE) and has improved our knowledge of their extent and evolution. Whereas the prognostic value of cranial ultrasound scan and of computed tomography is quite limited,3 recent publications have shown the value of MRI for predicting prognosis.4–7
Several studies have reported the use of the electroencephalogram (EEG) in HIE not only for the recognition of seizures8,9 but also as a reliable predictor of neurodevelopmental outcome. In particular, it has been demonstrated that it is the background activity, which best predicts the outcome,10–18whereas the significance of paroxysmal abnormalities19,20 is more controversial. In other studies brain electrical activity has been evaluated by means of an amplitude-integrated EEG, which is also a good predictor of outcome.21–24
Less is known about the correlation between specific EEG patterns and specific brain lesions or on whether the combined use of these 2 techniques can improve our prognostic capability in infants with HIE. This study investigated the relationship between different EEG abnormalities and specific patterns of brain lesions seen on MRI scan and their prognostic value in a cohort of neonates with HIE.
MATERIALS AND METHODS
The infants described in this study are part of a large cohort of full-term infants with hypoxic-ischemic brain injury born at or referred to the Hammersmith Hospital. Ethical permission for this study was obtained from the Hammersmith Hospital Research Ethics Committee. Parental consent was also obtained in all cases. The diagnosis of acute neonatal encephalopathy was made in infants who showed signs of fetal distress before delivery (including abnormal cardiotocograph recordings such as decreased variability, late decelerations, and a baseline bradycardia), who had abnormal Apgar scores (<5 at 1 minute and <7 at 5 minutes), requiring resuscitation at birth and who developed specific neurologic abnormalities during the first 24 hours after delivery. Term infants (>37 weeks' gestational age) were included in the study if they fulfilled the criteria for the diagnosis of acute encephalopathy and had at least 1 EEG in the first 72 hours after birth and 1 neonatal brain MRI after the first week.
EEGs were performed by means of a cassette computer-aided apparatus (Oxford-Medilog) and included 4 or 8 EEG and polygraphic leads: at least 2 EEG bipolar leads (C4-O2 and C3-O1 of the International 10–20 System) and pneumogram were always included. Most recordings, however, had more extended EEG montages (Fp2-T4, C4-O2, Fp1-T3, C3-O1), pneumogram and other polygraphic leads (electrocardiogram, electroculogram, etc.). EEGs lasted 6 to 24 hours and were examined by a neurophysiologist (E.B.) blinded to the MRI and outcome data.
In each EEG the occurrence of background and paroxysmal abnormalities was evaluated.
Three different types of background abnormalities were observed: constant low voltage, constant discontinuity, and dysmaturity.
A tracing was defined as constant low voltage when the detected background activity amplitude was constantly <20 μV.
Constantly discontinuous tracings, so defined when there was a constant alternating of relatively high-amplitude bursts and low voltage (at least <45 μV) intervals, were classified according to the following criteria18:
• extreme discontinuity: maximum interval duration >40 seconds;
• severe discontinuity: maximum interval duration 20 to 40 seconds;
• moderate discontinuity: maximum interval duration <20 seconds
A tracing was classified as dysmature when the observed maturational features were not appropriate for the postmenstrual age of the patient, resembling those of a younger infant. The established limit for considering an EEG as dysmature was a difference of >2 weeks.25–27
The incidence of paroxysmal abnormalities (ie, abnormal EEG transients) was evaluated according to previously published criteria.20 In particular, sharp waves, alpha discharges and sharp rhythmic delta and theta activities were considered all together (eg, the occurrence of a sharp wave was counted as the occurrence of a rhythmic sharp theta activity) and their incidence was scored according to the following classification:
• absent: no abnormal transients detected;
• very rare: <1 abnormal transient per 5 minutes;
• rare: at least 1 abnormal transient per 5 minutes but <1 abnormal transient per 1 minute;
• moderately frequent: at least 1 abnormal transient per 1 minute but <1 abnormal transient per 10 seconds;
• frequent: at least 1 abnormal transient per 10 seconds
A train of sharp waves (or an alpha discharge or a rhythmic sharp theta or delta activity) was considered as interictal when lasting <5 seconds and ictal when lasting >5 seconds,9irrespective of possible clinical correlates.
When ictal EEG discharges were present, they were classified as follows:
• rare: when they occupied less than 5% of the recording;
• moderately frequent: when they occupied 5% to 30% of the recording;
• status epilepticus: when they occupied more than 30% of the recording.
MRI was performed using a 1 Tesla HPQ magnet (Marconi Medical System, Cleveland, OH). All the infants had early and serial MRI. Only the images obtained after the first week after birth (range: 7–28 days) were considered in this study as 1) by this time brain swelling, usually present in the first days after birth, has cleared and the pattern of lesions is more evident and 2) very early scans can appear relatively normal even with severe insults. Images were obtained in the transverse plane with T1-weighted spin echo ([SE] 860/20), T2-weighted SE (3000/120) and age-related inversion recovery ([IR] 3800/30/950) sequences. Images were assessed for abnormal signal intensities by an experienced observer (M.A.R.), blinded to EEG and outcome data. The pattern of abnormal signal intensities observed was documented as follows.
The posterior limb of the internal capsule (PLIC) was assessed as normal, equivocal, or abnormal according to our previously published criteria.6
The basal ganglia and thalami were assessed as normal, minimal, moderate, and severe. Minimal: Focal abnormalities but normal signal within the PLIC. Moderate: Focal abnormalities involving the posterior lentiform nuclei and ventrolateral nuclei of the thalami with equivocal or abnormal signal intensity within the PLIC. Severe: Widespread abnormalities in all regions of the basal ganglia and thalami and abnormal signal intensity within the PLIC.
White matter abnormalities were documented according to which lobes of the brain were involved, whether there was a hemorrhagic element to the lesion and whether they were subcortical, periventricular, or widespread. In some neonates minimal changes of long T1 and long T2 in the periventricular white matter are difficult to differentiate from normal appearances and for the purposes of the present study these were not classified as abnormal. Abnormalities in the white matter were described as moderate or severe. Moderate: small focal lesions with a short T1 and short T2, consistent with hemorrhage and/or areas with an exaggerated long T1 and long T2 but no loss of gray/white matter differentiation. Severe: more marked areas of abnormality with larger hemorrhages or exaggerated long T1 and T2 with loss of gray/white matter differentiation, consistent with infarction.
The cortical abnormalities consisted of highlighting with an abnormally high signal on T1-weighted images and were graded according to how many cortical sites were involved.5
The scans were classified according to the predominant pattern observed into:
• normal: normal basal ganglia and thalami, white matter and cortex. This group included infants who may have minimal periventricular white matter change of prolonged T1. All the infants in this group had shown brain swelling on the early scans performed in the first days of life.
• minimal basal ganglia and thalami: focal abnormalities in the basal ganglia and thalami, normal PLIC and normal white matter and cortex.
• moderate white matter: focal abnormalities in the white matter with or without cortical involvement but with normal basal ganglia, thalami, and PLIC.
• moderate basal ganglia and thalami: focal abnormalities in the basal ganglia and thalami, equivocal or abnormal PLIC, with or without cortical highlighting.
• moderate white matter and basal ganglia and thalami: focal abnormalities in the white matter and mild or moderate abnormalities in the basal ganglia and thalami with or without cortical involvement.
• severe white matter: multifocal abnormalities with or without white matter hemorrhage with normal basal ganglia and thalami and PLIC.
• severe basal ganglia and thalami with subcortical white matter: widespread abnormalities in the basal ganglia and thalami always with abnormal PLIC with focal abnormalities in the subcortical white matter and in the cortex.
• severe basal ganglia and thalami with diffuse white matter: widespread abnormalities in the basal ganglia and thalami with abnormal PLIC with widespread abnormalities in the white matter and cortex.
This was evaluated using a standardized protocol of neurologic examination28,29and Griffiths' developmental scales.30
The outcome was classified as normal or abnormal at the age of 2 years. Infants were classified as normal if they had no abnormal neurologic signs and a developmental quotient (DQ) >85. Infants with DQ between 75 and 84 and/or with mild neurologic signs, but who were able to walk independently were classified as mildly abnormal. Infants with DQ between 50 and 74 and/or cerebral palsy, who were able to sit without support but not to walk were classified as moderately abnormal. Infants with DQ <50 and/or cerebral palsy who were not able to sit unsupported were classified as severely abnormal. This group also included the infants who died, in the first weeks or months of life.
Background EEG activity was normal in 8 infants. One EEG was classified as dysmature. Constant low voltage was observed in two infants and constant discontinuity was observed in 14 (moderate in 3, severe in 8, and extreme in 3).
Abnormal EEG transients were detected in all tracings but 1. Within each recording their incidence ranged between rare and frequent. EEG discharges were observed in 13/25 patients: they were rare in 6 traces and moderately frequent in 3. Status epilepticus was observed in the remaining 3 cases. Both abnormal EEG transients and EEG discharges were seen more often in constantly discontinuous tracings and, in particular, more in severely discontinuous than in moderately or extremely discontinuous ones.
Five of the 25 infants had normal scans by the end of the first week and did not show any sign of brain swelling which was invariably present on the scan performed in the first days.
One case showed minimal and 3 moderate abnormalities in the basal ganglia and thalami. Three infants had moderate and 4 had severe white matter abnormalities associated with a variable degree of cortical highlighting but with preserved basal ganglia.
Two infants showed moderate white matter and basal ganglia abnormalities.
Seven infants showed severe basal ganglia lesions, associated with subcortical white matter abnormalities in 2 and with diffuse white matter changes in the other 5.
Eight infants had a normal outcome, 1 had a mild and 7 a moderately abnormal outcome. Four infants developed a severely abnormal outcome and 5 died.
Correlation Between EEG Findings and Outcome
Table 2 and Graph 1 show details of the correlation. All but 1 of the 8 infants with normal EEG background activity had a normal outcome. The case with a dysmature tracing also had a normal outcome. All 14 infants with constantly discontinuous EEG showed an abnormal outcome. One of 2 infants with a constant low voltage tracing died in the neonatal period and the other showed a severely abnormal outcome.
The incidence of abnormal transients and the occurrence of EEG discharges did not seem to relate to the outcome.
Correlation Between MRI Findings and Outcome
The 5 infants with normal MRI, the 1 with minimal basal ganglia and thalami abnormalities and 2 of 3 with moderate white matter abnormalities all had a normal outcome. The 3 infants with moderate basal ganglia abnormalities and all 4 with severely abnormal white matter had a moderately abnormal outcome. All the patients with moderate white matter and basal ganglia involvement and the ones with severe basal ganglia and white matter involvement had a severely neurodevelopmental outcome or died. Table 2 and Graph 1 show details of the severity of the outcome.
Correlation Between EEG and MRI Findings and Outcome
All infants with a normal MRI, with minimal basal ganglia, and thalami lesions and 1 of 3 with moderate white matter lesions had a normal EEG background activity. Another infant with moderate white matter abnormalities had a dysmature EEG (Fig 1A, B). The outcome was normal in all cases.
Three infants had moderate basal ganglia and thalami lesions, background EEG was normal in 1, and discontinuous in the remaining 2. The outcome was abnormal in all 3.
All 4 infants with severe white matter lesions had severely discontinuous background EEG and moderate outcome abnormalities (Fig 2 A, B).
All patients with moderate white matter and basal ganglia lesions, and severe basal ganglia and white matter lesions had a discontinuous or low voltage EEG (Fig 3 A, B). All had a severely abnormal outcome.
The aim of this study was to evaluate the correlation between early EEG and neonatal brain MRI findings in term infants with neonatal acute encephalopathy and hence determine their relative merits in this clinical situation.
We choose to analyze the EEG performed within 3 days after birth as its diagnostic and prognostic accuracy is best soon after an insult.12 It is well-established that an early EEG has very good prognostic value in full-term infants with HIE. Constantly discontinuous patterns, sometimes defined as burst-suppression, have already been related to an unfavorable outcome.10–18 This applies not only to the full EEG but also to EEG recordings from a limited number of leads17 and also to the amplitude integrated EEG.21–24
In contrast, we only analyzed the MRI scans performed at the end of the first week after birth and before 1 month. Imaging techniques do not reflect the full extent of injury for some days. The pattern of injury is most easily seen between 1 and 4 weeks after birth, when brain swelling has settled, and before atrophy becomes obvious. It has been reported that this is the best time for predicting both the type and severity of outcome.6
We were able to demonstrate a good correlation between EEG and MRI findings. Normal MRI scans and minimal basal ganglia lesions were always associated with a normal EEG background, while EEG background was always abnormal when the MRI showed severe lesions such as severe basal ganglia and/or white matter lesions. The most abnormal EEG findings (extreme discontinuity, low voltage) were only found in patients with the most severe MRI abnormalities (ie, severe basal ganglia and thalami with diffuse or subcortical white matter lesions). In contrast moderate white matter and/or basal ganglia lesions were associated with variable degrees of continuity of background activity.
There was only 1 child in whom there was an obvious discrepancy between the EEG and the MRI. This was an infant with very focal basal ganglia and thalamic lesions who had a normal background EEG. The other 2 infants with this type of lesion had an abnormal EEG.
The 2 techniques were both highly predictive of outcome. Even with a reduced number of leads, the early EEG background activity was a reliable predictor of normal or abnormal outcome. All but 1 of the infants with a normal EEG background had a normal outcome. The exception was 1 of the 3 infants with very focal basal ganglia and thalamic lesions. All 3 of these infants had a moderately abnormal motor outcome with good preservation of head growth and intellect. We do not know why the EEG did not reflect the injury in 1 infant. They all had similar degrees of cortical involvement around the central sulcus and all had a similar pattern of central damage. Although this was the only exception it did constitute 1 in 9 infants with a normal EEG. It is, therefore, important to recognize that this particular pattern of damage, which is followed by a significant motor deficit, may occasionally occur with a normal EEG.
All infants with a discontinuous EEG had an abnormal outcome and those with an extremely discontinuous EEG or low voltage died or had severely abnormal outcome. However, moderate and severe discontinuity were associated with various degrees of severity of abnormal outcome, ranging from mild to severe abnormalities.
Dysmaturity was only found in 1 infant and was associated with a normal outcome, confirming it as a minor EEG abnormality when observed at term age.11,26 In a previous study18 moderate discontinuity was sometimes followed by normal neuromotor development: this was not confirmed in our cohort, possibly because of the small numbers. Abnormal EEG transients and discharges were found in the majority of the children but their frequency within a trace was relatively rare both in infants with normal background and good outcome and in those with very severe EEG background abnormalities. They were not predictive of outcome are these observations are consistent with a previous study showing the poor prognostic value of EEG transients.20 The low frequency of EEG transient and discharges within each trace might be partly explained by the fact that all the infants with HIE 2, who had convulsions, were treated with phenobarbitone before the EEG recordings were started.
The MRI performed at the end of the first week of life gave more specific information on the type and the severity of sequelae developed by these children than the EEG. All the infants with normal scans, mild basal ganglia, or moderate white matter lesions had a normal outcome. Children with severe white matter changes developed cerebral palsy and become microcephalic but eventually acquired the ability to walk and only had a moderate global delay. The severity of the outcome in the other groups seemed to depend on a large degree on the extent of the basal ganglia involvement. Infants with more severe and diffuse basal ganglia lesions invariably developed cerebral palsy (often dystonic), severe global delay and microcephaly. In contrast, the children with more discrete basal ganglia lesions developed less severe motor involvement, maintained normal head growth and had better cognitive development.
Our results show that a combination of early EEG and slightly later neonatal MRI scans provides an important contribution to the management of infants with HIE. The early EEG distinguishes between those infants with a normal or abnormal outcome and can be performed at the bedside in the neonatal unit. If it is abnormal, proceeding to the MRI after 1 week will give more specific information about the type of outcome to be expected.
We thank Dr D. Wertheim, D. Murdoch Eaton, and the late Cindy Bradshaw for their help in recording and interpreting the EEGs.
- Received March 27, 2000.
- Accepted July 27, 2000.
Reprint requests to (E.M.) Department of Paediatrics and Neonatal Medicine, Imperial College School of Medicine, Hammersmith Hospital, Du Cane Rd, London W12 0HN, United Kingdom. E mail:
- HIE =
- hypoxic-ischemic encephalopathy •
- MRI =
- magnetic resonance imaging •
- HIE =
- hypoxic-ischemic encephalopathy •
- EEG =
- electroencephalogram •
- SE =
- spin echo •
- IR =
- inversion recovery •
- PLIC =
- posterior limb of the internal capsule •
- DQ =
- development quotient
- Volpe JJ. Neurology of the Newborn. 3rd ed. Philadelphia, PA: WB Saunders; 1995
- Rutherford MA,
- Pennock JM,
- Counsell SJ,
- et al.
- Barkovich AJ,
- Hajnal BL,
- Vigneron D,
- et al.
- Mizrahi EM,
- Kellaway P
- Biagioni E,
- Bartalena L,
- Boldrini A,
- Pieri R,
- Cioni G
- al Naqeeb N,
- Edwards AD,
- Cowan FM,
- Azzopardi D
- Stockard-Pope JE, Werner SS, Bickford RG. Atlas of Neonatal Electroencephalography. 2nd ed. New York, NY: Raven Press; 1992
- Dubowitz LMS, Dubowitz V, Mercuri E. The Neurological Assessment of the Preterm and Full-Term Infant. Clinics in Developmental Medicine. London, United Kingdom: McKeith Press; 1998. No. 148
- Griffiths R. The Abilities of Babies. London, United Kingdom: University of London Press; 1954
- Copyright © 2001 American Academy of Pediatrics