Lawrence Buadu, MD PhD, Sven Ekholm MD PhD, Ann Lenane MD, Toshio Moritani MD, Akio Hiwatashi MD, PL Westesson MD.
University of Rochester Medical Center, Rochester, New York
Scope of the problem
It is estimated that more than 2000 children in the United States die each year as a result of child abuse . Many think that the number is actually higher because many child fatalities, that are actually related to abuse are reported as accidents, homicides or sudden infant death syndrome. Nonaccidental head injury (NAHI) is largely restricted to children under three years of age, with the majority occurring during the first year of life . Inflicted head injury is the most common cause of traumatic death in infancy [2, 3]. With inflicted head injury an accurate history is rarely provided at presentation. The history provided may be vague or may vary with time . Physical examination, although useful may provide little insight regarding underlying brain injury. Consequently, the diagnosis and detection of nonaccidental head injury (NAHI) usually comes to rest on radiologic imaging. If radiographic indicators of abuse or neglect are missed, it portends grave consequences for the child who will invariably be returned to a high risk environment. It is therefore crucial for radiologists to be familiar with the imaging findings of NAHI.
1. Review the common radiological features of NAHI.
2. Highlight the subtle and less apparent indicators of NAHI.
Biomechanics and Terminology
The various terminologies applied to inflicted head injury in children reflect the evolution in our understanding of the underlying mechanisms necessary to cause some of the injuries seen. The term “whiplash shaken baby syndrome” was originally coined by Caffey to explain the constellation of findings of subdural and subarachnoid hemorrhages, traction type metaphyseal fractures and retinal hemorrhages in children . Since then terms like shaken baby syndrome, shaken impact syndrome and shaken infant syndrome have all been used in an attempt to explain underlying mechanisms of inflicted head injury infants. Regardless of terminology it is well accepted that most inflicted head injuries in children are of the dynamic type. Dynamic injuries may occur in either direct contact trauma or indirect injury. Contact phenomena result in localized distortion or a fracture of the skull, a focal cortical injury, epidural hematoma or subdural hematoma. In contrast to direct trauma, indirect injuries are independent of skull deformation and entail inertial loading which occurs with sudden acceleration or deceleration of the head . Although a contact may occur with this mechanism, significant life threatening injuries may occur without an impact. Head acceleration or decelerations results in a variety of strain deformations of the skull and its contents. Shear strain deformation, which produces disruption at tissue interfaces is the most important mechanism in the production of intracranial injury. Furthermore the primary injury occurring with these biomechanical forces may result in other pathophysiologic alterations or secondary injury (e.g. edema, swelling, hypoxic ischemia, herniation) and produce additional imaging findings.
Scalp Injury (Fig 1)
Scalp injuries are usually the result of direct impact but may not be apparent in inflicted head injuries. When present, these may manifest as abrasion, bruising, laceration, or a burn; subcutaneous hemorrhage or edema (caput succedaneum); subgaleal hemorrhage or a subperiosteal hemorrhage (cephalhematoma). Although CT is well suited to the evaluation of these fluid collections, MR imaging with its superior soft tissue resolution shows these changes to better advantage (fig 1a & b)
Cranial Injury (Fig 2)
The prevalence of skull fractures in all cases of abuse is 10% to 13% . The radiographic appearances of skull fractures may be classified into simple and complex categories. CT scan may show a linear defect on axial sections with bone algorithm (fig 2a) however if the fracture is in the plane of the scan it can easily be overlooked. 3D reconstructions are helpful but may obscure the fracture line due to a smoothing effect (fig 2c). Maximum intensity projection images (MIP) are especially sensitive and depict fractures to best advantage (fig 2d).
Extra-axial (Figs 3,4)
Extra-axial lesions are usually hemorrhagic in nature. Hemorrhage can be epidural, subdural or subarachnoid. Epidural hematomas (EDH) are infrequently encountered in infancy and are particularly uncommon in cases of abuse. They are usually the result of direct impact injuries and are often associated with skull fractures . In contrast, nonaccidental subdural hemorrhage is much more common, usually caused by high energy, angular or rotational acceleration deceleration forces delivered during shaking or shaking impact assaults. Shear strain forces result in disruption of delicate cortical bridging veins as they leave the cortical surfaces to enter the dural venous sinuses. The injury most frequently involves the cortical venous structures draining into the superior sagittal sinus. Consequently, the smallest and earliest collections are encountered in the interhemispheric regions over the cerebral cortices. Because the underlying mechanisms are similar there is a high association of retinal hemorrhages in nonaccidental trauma with SDH. Acute subdural collections are hyperdense on CT. However, subacute or chronic subdural hematomas tend to be of low or mixed attenuation on CT and are better delineated on MR imaging (fig 3a & b). The mechanism of subarachnoid hemorrhage (SAH) is similar to that of SDH resulting from the disruption of cortical veins occurring with angular accelerations or decelerations of the head. In contrast to its high sensitivity for detecting SDH, MRI is relatively insensitive to the presence of hyperacute or acute SAH. Fluid attenuated Inversion Recovery (FLAIR) imaging is, however, quite sensitive and has resulted in improved detection of SAH (fig 4).
Most intra-axial lesions in contrast to extra-axial lesions are nonhemorrhagic although hemorrhagic lesions can occur (fig 5). Nonhemorrhagic intra-axial lesions which are more difficult to identify early in their course and are responsible for most deaths from inflicted head injury . Nonhemorrhagic intra-axial lesions may present as diffuse pathologic alterations like hyperemic cerebral swelling, diffuse cerebral edema or hypoxic ischemic injury. More focal manifestations include focal infarcts or axonal injury.
Diffuse Cerebral Edema
Brain edema is the most profound pathologic alteration encountered with inflicted brain injury, yet the most poorly understood. Brain edema a consequence of increased brain water (cytotoxic and vasogenic) may occur as a response to direct focal injury such as cerebral contusion or diffuse primary injury such as diffuse axonal injury (DAI). Furthermore, vascular occlusion due to cerebral brain stem herniation as well as pressure necrosis may lead to cerebral edema. CT images obtained immediately after the traumatic event often show no evidence of swelling or edema. Swelling or edema may become manifest on CT within a few hours with extensive loss of gray-white differentiation and diffuse hypodensity. These findings carry a poor clinical outcome regardless of the clinical grade.
Hypoxic Ischemic Injury (Fig 6)
There is a tendency for profoundly injured infants to develop CT manifestations of brain edema that primarily involves the cerebral cortex and subcortical white matter but apparently spares the basal ganglia, thalami brainstem and cerebellum. This finding is often associated with subdural hematoma and can be unilateral or bilateral. Cohen and colleagues  coined the term reversal sign to describe this phenomenon (fig 6). The pathogenesis of the reversal sign is not entirely understood but experimental studies appear to indicate that cerebral cortical gray matter is particularly sensitive to hypoxic ischemic injury. Bird and associates suggest that the peripheral low density with relative central high density is related to the passive congestion and distension of deep medullary veins because of partial venous outflow form obstruction from the increased intracranial pressure .
Shear Injury (Fig 7)
Shear injury of the white matter generally referred to as diffuse axonal injury results from angular acceleration during shaking or blunt impact trauma. Histologically the lesion is characterized by axonal swelling or the so called retraction balls. Lesions are commonly noted in the cerebral hemispheres at the gray-white matter junctions (fig 7), the corpus callosum, the dorsolateral aspect of the upper brainstem, the upper pons and the basal ganglia. Because of the superior conspicuity provided, T2* gradient echo and FLAIR imaging are the preferred MRI techniques for demonstrating DAI. However, DAI is particularly uncommon in infants.
Atrophy (Fig 8)
Atrophy is often the result of primary and secondary traumatic brain injury. When serial imaging demonstrates an evolution from widespread cerebral edema to cerebral atrophy it is likely that hypoxic ischemia has played a major role in the cerebral injury. The time course for the development of cerebral atrophy is variable but the imaging findings may develop rapidly when the initial insult is severe (fig 8). These findings correspond with the development of cerebral spasticity and a vegetative state.
Despite the major advances made in recognizing indicators of NAHI, in some instances there may be no apparent morphological findings despite significant underlying brain injury. In these instances MR spectroscopy can be useful and is becoming an important part of the diagnostic armamentarium in helping to unmask biochemical alterations which may predate any morphological changes. Additionally MR spectroscopy has been shown to have prognostic implications in NAHI .
Discussion & Conclusions
Inflicted head injury is the most common cause of traumatic death in infancy, however history is often unreliable and physical exam may be unrevealing. Diagnostic imaging therefore plays a crucial role in identifying potential patterns of abuse. Although no single imaging finding is specific for abuse, no other medical condition fully mimics all the features of non-accidental injury in infants and children. As radiologists we have a vital role to play in identifying those imaging findings that can suggest abuse. Our index of suspicion should be high since failure to identify potential patterns of abuse portends grave consequences for the child who will invariably be returned to a high risk environment (fig 9). In this presentation we have attempted to demonstrate the common and some less common patterns of nonaccidental head injury which have been significantly enhanced since the introduction of MR imaging. Despite all the technological advances, however, imaging of nonaccidental injury continues to be a challenge. Some forms of injury like intermittent suffocation and asphyxiation may present with little or no morphological changes on imaging. MR spectroscopy, however, holds promise for the future by aiding in the identification of biochemical changes that may predate and morphological findings. This may help identify children who are subject to subclinical forms of repetitive abuse before a fatality occurs.
1. National Clearing house on Child Abuse and Neglect Information. (No date) Child fatalities fact sheet [online] Available: www.calib.com/nccanch/pubs/factsheet/fatality.html[2000, February 22]
2. Centers for Disease Control. Childhood injuries in the United States. Am J Dis Child 1990; 627-46.
3. Billmire ME, Myers PA. Serious head injury in infants; accident or abuse? Pediatrics 1985; 75: 340-2
4. Duhaime AC, Gennarelli TA, Thibault LE, Bruce DA, Margulies SS, Wiser R. The Shaken baby syndrome: A clinical, pathological and biomechanical study. J Neurosurg 1987; 66:409-15
5. Caffey J. The whiplash shaken infant syndrome; manual shaking by the extremities with whiplash-induced intracranial and intraocular bleedings, linked with residual permanent brain damage and mental retardation. Pediatrics 1974; 54: 396-403
6. Kleinman, Paul K. Diagnostic imaging of child abuse-2nd edition; pg 286-287.
7. James HE, Shut L: The neurosurgeon and the battered child, Surg Neurol 2:415-418, 1974
8. Merten DF, Osborne DRS: Craniocerebral trauma in the child abuse syndrome: radiological observations, Pediatr Radiol 14: 272-277, 1984
9. Kleinman, Paul K. Diagnostic imaging of child abuse-2nd edition; pg 296-297.
10. Cohen RA, KaufmanRA, Myers PA, Towbin RB: Cranial computed tomography in the abused child with head injury, AJR 146:97-102, 1986
11. Bird CR, Drayer BP, Gilles FH: Pathophysiology of "reverse" edema in global cerebral ischemia, AJNR 10: 95-98, 1989
12. Hasler LJ, Arcinue E, Danielsen ER, Bluml S, Ross BD. Evidence from proton magnetic resonance spectroscopy for a metabolic cascade of neuronal damage in shaken baby syndrome.
Fig 1. Scout image from a CT exam in an 8-month-old male with suspected NAHI head injury shows biparietal soft tissue swelling (fig 1a.). Coronal T1 gradient echo images (fig 1b) show the biparietal subgaleal hemato-mas to better advantage.
Fig 2. 8-month-old male with suspected NAHI (same patient as fig 1). Axial nonenhanced CT exam with bone algorithim shows a linear defect (arrow) in the right parietal region (fig 2a) consistent with frac-ture. A second fracture in the left parietal region (arrow) is less apparent. 3D recon-structed images (fig 2b) depict the right parietal fracture clearly (arrows) however, the left parietal fracture (arrows) is less apparent due to a smoothing effect (fig 2c). MIP images (fig 2d) shows the left parietal fracture to best advantage.
Fig 3. 5-month-old male child with nonreactive pupils and sus-pected NAHI. Coronal T1 SPGR image demonstrates a left interhemispheric SDH (fig 3a). A right SDH (arrows) which is less apparent on the TIWI is seen to better ad-vantage on the more sensitive gradient echo image (fig 3b).
Fig 4. 9-month-old male presenting with suspected nonaccidental trauma and retinal hemorrhages. Sagittal T1-weighted image shows a right SDH (fig 4a). Axial fluid attenuated inversion recovery image demonstrates SAH (arrows) in the right parietal region (fig 4b).
Fig 5. 5-month-old female who presented with altered mental status and retinal hemorrhages on physical examination. Sagittal T1-weighted (fig 5a.) and gradient echo (fig 5b) images demonstrate an intra-parenchymal hematoma.
Hypoxic Ischemic Injury
Fig 6. 2-month-old female infant presenting with seizures and apnea. Non-enhanced CT (NECT) shows the reversal sign with diffuse and extensive hypodensity of the cerebral cortices and relative sparing of the basal ganglia and cerebellum.
Fig 7. 14-day-old male infant with new onset focal seizures, fever and swelling on the right forehead. Axial CT shows a focus of hemorrhage over the left temporal tip (arrow) (fig 7a). Axial T1 and T2-weighted images confirm the presence of hemorrhage at the right temporal tip (fig 7b & c). Diffusion weighted image shows two punctuate foci of restricted diffusion (arrows) in the left parietal lobe most consistent with axonal injury (fig 7d). ADC values (not shown) were diminished.
Fig 8. 2-month-old female infant presenting with seizures and apnea (same patient as fig 7.) Initial midline sagittal T1WI shows relative preservation of parenchymal volume. Follow-up MR image 9 days later shows significant loss of volume and the relatively rapid progression of severe diffuse brain injury to atrophy.
Fig 9. 2-year-old female who initially presented with seizure. Axial T1-weighted image (fig 9a) shows a small right SDH (arrows). Axial flair image (fig 9b) shows a small amount of SAH (arrows). DWI was normal. Child injury survey(not shown) at the time was also normal. A month later the child returned with a history of a fall which resulted in a right tibia/fibula fracture. A repeat MR exam shows multiple areas of subacute infarction on T1, flair and DWI (fig 9d, e & f).