Simon R. Platt, BVM&S, MRCVS, DACVIM (Neurology), DECVN
Head trauma constitutes an important cause of morbidity and mortality in both humans and animals. Concerted efforts to improve the clinical outcome of each patient with head injury can only be effective if 1) the true extent of the initial brain injury can be defined, and 2) we develop an understanding of what this injury actually means in terms of the treatment required and the prognosis for the individual. At the present time, we are just coming to terms with the clinical indications of brain injury and its prognostic implications, through scoring systems such as the modified Glasgow Coma Scale (MGCS). In veterinary medicine, imaging has played a small subjective role in efforts to predict the final outcome after a head injury; however, with the increasing use of magnetic resonance imaging (MRI), we have the opportunity to develop objective and more accurate predictions of outcome.
In addition to a prognostic indicator, we can identify post-traumatic lesions which may require surgical therapy. Until our knowledge of these factors is much improved though, we must ask ourselves, of what benefit is this particular imaging modality going to be for this individual patient? This lecture briefly reviews the current imaging choices which we have available for the head trauma patient in veterinary medicine and their limitations as well as presents recent data assessing the prognostic value of MRI in head trauma patients.
In traumatic brain injury, the mechanism of damage can be classified as primary and secondary. In general, structural imaging techniques are used to visualize primary brain injury. Primary brain injury occurs at the moment of impact, with diffuse axonal injury being the most important primary lesion. Diffuse axonal injury is a consistent finding in mild, moderate, and severe traumatic head injury, although the severity increases with that of the head injury. Primary brain injury also comprises focal abnormalities, such as contusions and hematomas, as a result of either direct external contact forces or from the movement of the brain within the skull.Secondary brain injury, on the other hand, develops within hours after impact as a result of primary injury and mainly consists of ischemia; unfortunately this is best visualized using functional imaging which is rarely attempted in veterinary medicine even where available. In the emergency human setting, clinical management is guided by the imaging of structural abnormalities requiring acute interventions and during follow-up, structural imaging techniques are most commonly used to explain post-concussional symptoms or to predict outcome.
Imaging Techniques Available
Skull radiographs are unlikely to reveal clinically useful information about brain injury but may occasionally reveal evidence of calvarial fractures. The canine skull presents problems both in radiography and the interpretation. One major difficulty which needs to be taken into account is the obvious variability of the anatomy between breeds. Anesthesia is essential for accurate positioning, but this may not be possible in the acutely traumatized patient without a certain amount of risk. In those patients in which there is radiographically apparent bony damage, the most common finding is a linear defect in the bone with minimal displacement of fracture fragments. It is often difficult to differentiate vascular channels that create relatively smooth lucencies with sclerotic borders and branching patterns from the more irregular fracture lines. Skull fracture fragments may be depressed into the brain. This is most common with fractures of the dorsal and lateral portions of the skull. Fractures of the base of the skull, middle ear, and temporomandibular joints may be difficult to assess. If fractures of the osseous bullae are present, the fragments may be displaced into their lumens.
Computed tomography has been the preferred modality for imaging the head in cases of severe human head injury especially if a surgical intracranial lesion is a possibility. This should be suspected in animals, which demonstrate neurological deterioration even though they are treated with aggressive medical therapy. However, even patients with 'mild' head trauma can exhibit abnormalities on the CT scan and so the initial decision to image the patient's head should not be based on a single neurological examination.1
In humans with mild and moderate head injury, there is no agreement about routine CT scanning. There is substantial variation among institutions in the ordering of CT for patients with mild head injury, ranging from 16% to 74%.2 When all patients with mild-to-moderate head injury are scanned, the incidence of abnormal findings is about 15%, increasing to 50% when a CT scan is done in only those patients with neurological symptoms.3 The overall sensitivity of CT to abnormalities in acute head trauma is 63-75%.4
Magnetic Resonance Imaging
In humans, MR imaging has been shown to provide key information relevant to the prognosis based upon its ability to detect subtle parenchymal damage not evident on CT imaging.5 The best MR protocol for a specific acute head injury patient is one which can expediently detect intracranial hematomas, identify non-hemorrhagic forms of injury, and provide sufficient anatomic detail to classify lesions. Although the level of bone detail may be poor, especially when compared to CT, most traumatic lesions can be detected with MRI, if images are obtained in at least two planes using T1-weighted, fluid attenuated inversion recovery (FLAIR), T2-weighted, and T2*-weighted sequences. The scientific basis behind these sequences is well documented in most imaging texts. The multiplanar capability of MRI is an advantage over the use of CT in which alternative planes need to be reconstructed from the transverse, cross-sectional, plane which inherently loses detail. Without such a capability, small lesions (e.g., traumatic brainstem lesions) may be missed, due a phenomenon termed partial volume averaging, whereby overlying normal tissue within a 3mm thick slice may obscure an abnormality. Additionally, localization of a lesion, essential if surgical therapy is to be considered, is only possible when there at least two views. The visibility of a lesion on a particular sequence is influenced by a number of factors which include lesion size and location, presence and age of hemorrhage and the presence of edema. Therefore the timing after the trauma when the MR imaging is performed effects the characteristics of the lesions due to the oxidative change of hemaglobin (discussed below) and the creation of edema secondary to progressive ischemia.
Overall, FLAIR and T2-weighted scans are most sensitive for detecting traumatic lesions across the whole time course of injury. In a human study, T2-weighted scans detected 93% of non-hemorrhagic and 93% of hemorrhagic lesions as compared with T1-weighted scans (68% and 87%) and CTs (18% and 90%), respectively.5 MR was initially thought to be insensitive for detection of some hematomas; however, it has become apparent that MR is extremely sensitive to hemorrhage throughout all stages of its evolution. The timing after the onset of the hemorrhage effects the state of oxygenation of hemoglobin; in the initial fully oxygenated state of oxyhemoglobin, the hematoma does not produce a signal much different from that of surrounding brain tissue, making acute lesions difficult to see and interpret; invariably there is enough anatomic distortion and peri-lesional edema to allow their recognition. Within a short space of time, the oxyhemoglobin is converted to deoxyhemoglobin, lasting up to a week and is very visible on T2-weighted images. With more time, the conversion of the hematoma to a predominant methemoglobin content, the lesion becomes extremely visible on T1-weighted images. In the acute stage, FLAIR-weighted sequences are used for the detection of diffuse axonal injury, edema, and hemorrhage whereas in the subacute and chronic stages FLAIR-weighted sequences are mainly used for the detection of gliosis. Although about 80% of diffuse axonal injury lesions were thought to be non-hemorrhagic in nature, improved MRI techniques indicate that the proportion of hemorrhagic diffuse axonal injury lesions is in fact much greater than previously thought.
T2*-weighted gradient-recalled echo sequences enable visualization of hemosiderin deposits as a result of hemorrhage, possibly due to diffuse axonal injury. T2*-weighted gradient-recalled echo imaging is better than T1-weighted and T2-weighted spin-echo sequences in the detection of traumatic lesions. In a study of patients with head injury of varying severity, the number of lesions on T2*-weighted gradient-recalled echo sequences was positively correlated with outcome, whereas on T2-weighted imaging it was not.6
Diffusion-weighted imaging is another MRI modality that is primarily used to detect vasogenic or cytotoxic edema. This technique is sensitive to the random movement of water molecules and can distinguish between lesions with increased and restricted diffusion in patients with head injury. The apparent diffusion coefficient can be calculated and used to quantify the degree of restriction of water molecules caused by head injury. Diffusion-weighted imaging is widely used in cerebral ischemic stroke, showing changes before the onset of visible abnormalities seen on conventional imaging. In a few studies comprising mild head injury, diffusion abnormalities were seen within days of injury. In severe head injury, diffusion-weighted imaging can be used to identify diffuse axonal injury as hyperintense lesions that are not visible on T2-weighted spin-echo, T2*-weighted gradient-recalled echo, or FLAIR sequences. Most of these diffuse axonal injury lesions show decreased diffusion probably due to cytotoxic edema within daysto weeks of injury. Although cytotoxic edema predominates in head injury, there is high diffusion in the acute phase, probably due to vasogenic edema. Diffusion-weighted imaging is less sensitive than T2*-weighted gradient-recalled-echo images for detecting hemorrhagic diffuse axonal injury lesions.5 However, the volume of lesions depicted with diffusion-weighted imaging shows a stronger correlation with clinical outcome in patients with head injury than FLAIR, T2-weighted spin-echo, or T2*-weighted gradient-recalled-echo sequences.
The Canine Experience
In a study performed by the author, MRI assessment of head trauma was investigated for an association with severity of neurological dysfunction in dogs and as to whether it could be predictive of patient survival.7 Dogs presenting with evidence of head trauma and which were imaged with a 1.5T MRI were retrospectively evaluated. Criteria necessary for inclusion in the study were 1) imaging performed within 7 days of the trauma, 2) neurological examination at time of MRI enabling a modified Glasgow coma score (MGCS) to be estimated, 3) survival at 1 and 6 months after MRI, 4) T1, T2, T2* gradient echo and FLAIR weighted images. All images were blindly evaluated for 1) extra-axial hemorrhage, 2) intra-axial hemorrhage, 3) fractures (linear, comminuted, compound and/or depressed), 4) degree of parenchymal shift (mm), 5) single or multiple parenchymal lesions and, 6) MRI grade of severity (I-IV) modified from established criteria in human head trauma imaging.
The MRI parameters were individually evaluated with the estimated MGCS and survival at 1 and 6 months. A linear trend was demonstrated between survival at 1 and 6 months and the MGCS. Nineteen of 32 dogs (59%) had an abnormal MRI which was not associated with outcome at 1 or 6 months. MRI grade was significantly associated with outcome at 1 and 6 months; a higher class was associated with a reduced probability of survival. The presence of intra- or extra-axial hemorrhage was not associated with outcome at 1 or 6 months. Dogs with no midline shift were more likely to survive to 1 month than dogs with any evidence of midline shift. Whether a dog had a skull fracture or not and type of fracture were not associated with survival. The MGCS was not associated with the presence of extra-axial hemorrhage, skull fractures or fracture types. There were significant associations between the MGCS and abnormal MRI, MRI grade, presence of intra-axial hemorrhage and the degree of midline shift.
1. Platt SR, et al. Vet Rec 2002;151(8):243.
2. Stiell IG, Ann Emerg Med 1997 30;. 14.
3. Gomez P, et al, Br J Neurosurg 1996;10:453.
4. Naalt van der. J Ann Neurol 1999;46:70.
5. Metting Z, et al. Lancet Neurology 2007;6:699.
6. Yanagawa Y, et al. J Trauma 2000;49:272.
7. Platt SR, et al. J Vet Int Med 2007;21(5):1145.