Traumatic brain injury (TBI) is one of the most common neurological conditions and is associated with mortality in dogs [123456]. Since accurate assessment of TBI patient can facilitate appropriate treatment and yield better outcomes with prognostic information, various clinical assessment methods for TBI have been studied in humans and dogs [13789101112131415]. Diagnostic imaging modalities such as computed tomography (CT) and magnetic resonance imaging (MRI) are widely used in TBI patients because they can identify treatable injuries to prevent secondary damage and provide useful prognostic information [1278111516]. In human medicine, CT is the modality of initial choice for acute TBI patients because it is a rapid, widely available, and highly accurate method for the detection of intracranial hemorrhage or skull fracture [78101112]. However, in the subacute and chronic phase of TBI, MRI is preferred over CT because of its superiority in detecting non-hemorrhagic lesions, brainstem injury, edema, and diffuse axonal injury [78101116].

Susceptibility-weighted imaging (SWI), a relatively new MRI sequence, has recently gained prominence due to the increasing need for accurate diagnosis in TBI patients. Previous studies have reported that this new sequence type is more sensitive in detecting traumatic lesions than CT or conventional MRI, especially in cases involving cerebral microbleeds [710121416]. In veterinary medicine, there is a study elucidating the correlation between conventional MRI results and clinical status or prognosis in dogs with TBI [1]. However, to the best of the authors' knowledge, there are no reports on the use of SWI in dogs with TBI or on comparisons of conventional MRI and SWI, even though the use of SWI has been apparently increasing. The aims of this study were 1) to describe conventional MRI and SWI results for TBI lesion evaluation, and 2) to determine correlations between clinical status and conventional MRI or SWI results. Consequently, our study aimed to identify the appropriate MRI sequences to use in the assessment of dogs with TBI.

The medical records of all dogs presenting between April 2013 and May 2015 were reviewed. Inclusion criteria were as follows: 1) TBI based on clinical history, 2) no known history of intracranial central nervous system disease unrelated to the TBI, 3) was presented at the animal clinic center within 24 h of TBI, 4) Modified Glasgow coma scale (MGCS) score [5] recorded at presentation, 5) hospitalized for at least 3 days after the TBI, 6) 1.5 Tesla MRI performed in the first 5 days after TBI, and 7) at least four MRI sequences obtained, i.e., T1-weighted (T1W), T2-weighted (T2W), fluid-attenuated inversion recovery (FLAIR), and SWI. All animals were treated with a general treatment protocol in accordance with ethical requirements.

MRI scans were performed with a 1.5-Tesla magnet (Siemens, Germany). For MRI examinations, anesthesia was induced in each patient with intravenous propofol (Myungmoon Pharm. Co., Korea) at a dose of 8 mg/kg body weight and maintained with isoflurane (Hana Pharm. Co., Korea) and oxygen. An extremity matrix coil was used, and the patients were placed in ventral recumbency. The T1W, T2W, FLAIR, and SWI sequences were acquired in transverse planes on each patient (Table 1).

All images included in this study were uploaded to a DICOM-viewing station (Infinitt Healthcare Co., Korea). Location, intensity of the lesion on T1W, T2W, and FLAIR, existence of median shift, and skull fracture were assessed by two clinicians experienced in veterinary diagnostic imaging (D. Noh and K. Lee) on the basis of consensus. Lesion locations were categorized as 1) subdural; 2) frontal, temporal, parietal, or occipital lobe; 3) thalamus; 4) cerebellum; and 5) brainstem. Signal intensity in comparison to the adjacent brain tissue was evaluated as hypointense, isointense, or hyperintense on T1W, T2W, and FLAIR, respectively.

All lesions detected on conventional MRI and SWI were evaluated based on a previously reported modified MRI grading system (MRGr; Grade 1, normal brain parenchyma; Grade 2, lesions only affect the cerebral hemisphere, cerebellar parenchyma, or both without median shift; Grade 3, same as Grade 2 but with median shift; Grade 4, lesions affecting the corpus callosum, thalamus, or basal nuclei with or without lesions of a lower grade; Grade 5, unilateral brain lesions in the brainstem with or without lesions of a lower grade; Grade 6, bilateral brain lesions in the brainstem with or without lesions of a lower grade) [117].

Statistical analyses were performed using commercial software (IBM SPSS statistics, USA). Spearman correlation analysis was used to assess the correlation of MGCS with MRGr on conventional MRI, and MGCS with MRGr on SWI. The p values lower than 0.05 were considered statistically significant.

Eleven dogs met the inclusion criteria. Patient history and signalment are summarized in Table 2. All dogs were small-breed, including seven Maltese and one each of Dachshund, Pomeranian, Toy Poodle, and Yorkshire terrier. The median age at presentation was 4 years (range, 1–7 years). The most common cause of TBI was a fall in five dogs, followed by blunt trauma in four dogs and motor vehicle accident in two dogs. The clinical signs were recumbency (n = 7), semicomatose condition (n = 5), seizure (n = 4), head tilt (n = 3), nystagmus (n = 3), circling (n = 2), ataxia (n = 1), and hemiparesis (n = 1). All dogs were hospitalized and received standard treatment to obtain stabilization after TBI. Follow-up assessments were performed by telephone for eight dogs. Seven dogs recovered over time and three showed full recovery on follow-up examinations (Table 2). Dogs with a higher MGCS score and a lower MRGr showed a tendency to recover quickly, while one dog with an MGCS score of 12 and MRGr 5 was euthanized because of its severe clinical signs.

Conventional MRI revealed 23 lesions in 11 dogs. The locations of the lesions are summarized in Table 3. On T2W, these lesions were hyperintense (21/23), isointense (1/23), and hypointense (1/23). On T1W, the lesions were hypointense (16/23), hypo- to isointense (5/23), and hyperintense (2/23). On FLAIR, the lesions were hyperintense (20/23), iso- to hyperintense (1/23), isointense (1/23), and hypointense (1/23). Three lesions showed a central lesion that was hypo- to isointense on T2W, hypointense (2/3) or hyperintense (1/3) on T1W, and hypo- to isointense on FLAIR. SWI showed comparable findings for all 23 lesions, but it also detected numerous additional lesions in four dogs (Table 3). In particular, in one dog, conventional MRI revealed only one ambiguous thalamic lesion whereas SWI showed multifocal lesions throughout the brain parenchyma, including the cerebral lobe, cerebellum, and caudate nucleus. However, subdural hemorrhage was only detected on conventional MRI, not on SWI (Fig. 1). Skull fractures were detected in three dogs (Table 3).

The median MRGr was 2 both on conventional MRI and SWI (range, 2–5; respectively). There was a significant negative correlation between MRGr on conventional MRI and MGCS score (r = −0.685, p < 0.05). However, there was no significant correlation between MRGr on SWI and MGCS score.

We described the use of conventional MRI and SWI for the evaluation of TBI lesions in dogs; this is the first veterinary study using SWI in dogs with TBI. According to previous MRI studies in humans, the T2W sequence can detect edema, contusions, and lesions near the skull base and at the under surface of the cerebral lobe [810]. It also can be used to estimate the age of hemorrhage although it shows low sensitivity for intraparenchymal hemorrhage [10]. FLAIR is superior to T2W for detection of periventricular and superficial cortical lesions and diffuse axonal injury, even though it requires a long acquisition time [81011]. The T2^*-weighted gradient-echo MRI sequence can depict paramagnetic deoxyhemoglobin, methemoglobin, or hemosiderin in lesions and some calcifications. However, it cannot distinguish between hemorrhage and calcification [18]. In contrast to these scenarios, SWI, a fully velocity-compensated high-resolution three-dimensional gradient-echo sequence, allows improved detection of paramagnetic blood products, extravascular deoxyhemoglobin, and methemoglobin, based on differences in their magnetic susceptibility [81216181920]. It can also distinguish a paramagnetic substance from calcification and aids in the detection of small brain parenchymal hemorrhage with a well visualized venous structure, showing higher detectability than that of T2^*-weighted gradient-echo sequences [410121821]. Because of its characteristics, SWI has been used widely in human clinical assessments involving head trauma patients.

In this study, similar to the findings in human patients, SWI showed a superior ability to detect brain parenchymal TBI lesions with more clarity and less ambiguous aspects than that from conventional MRI. The detection of lesions only on SWI may be attributable to the above-mentioned superiority of SWI for detecting small brain parenchymal hemorrhage, which is categorized as a punctate hemorrhagic lesion (< 10 mm) in human patients, and it may indicate the presence of axonal injury [101221]. Furthermore, one dog with only one ambiguous thalamic lesion on conventional MRI showed multifocal lesions throughout the brain parenchyma on SWI. This result may further indicate the usefulness of SWI in dogs with TBI, especially in cases that are unexplainable on conventional MRI.

Even though the SWI is purported to be the most sensitive sequence for detecting hemorrhagic lesions, subdural hemorrhage was only detected on conventional MRI, not on SWI [10]. The subdural region is the narrow space interposed between the arachnoid and pia mater and adjacent to skull bone. Because of its anatomical structure, this region may show magnetic susceptibility artifacts due to the skull base–air interface on SWI, resulting in an undetected lesion. However, on conventional MRI, the presence of this anatomical structure may not affect the detection due to the superior ability of this technique to distinguish lesion from bone. This assumption is consistent with that in previous human study that showed the ability of T2W to detect lesions near the skull base [810]. That study was also in agreement with an earlier study that showed better detection of central brain areas than peripheral brain areas on SWI [22]. These results support the necessity and superiority of conventional MRI in cases of TBI patients with subdural hemorrhage.

In this study, the MRGr for conventional MRI was negatively correlated with MGCS scores, and dogs with a lower MRGr tended to recover quickly. The MGCS, which evaluates clinical status, including motor activity, brainstem reflexes, and consciousness, has served as a useful prognostic index for dogs with TBI [135]. Although the Glasgow coma outcome scale was not analyzed in our study, the results suggest that conventional MRI could be used for predicting a prognosis, which is in agreement with other reports in dogs and humans [1]. However, there was no correlation between the MRGr on SWI and MGCS score. There are several possible reasons for this result: 1) Early clinical status was unaffected by the additionally detected small lesions, 2) MRGr was an incongruent method to adjust on SWI, and 3) only a small number of dogs with a narrow range of MRGr were included in this study. Previous human studies have shown a correlation between Glasgow coma scale scores and SWI findings by using the number of punctate lesions on SWI [1014]. This implies that SWI can reflect the early clinical status whereas our results could not. Our result may be attributed to the narrow range within the MRGr, even though it has been used as an evaluation method for head injury. We also could not exclude the possibility that the MRGr may not be an appropriate method for evaluating punctate lesions. To elucidate a possible correlation between clinical status and SWI findings, further studies using variable methods and a large number of dogs are needed.

This study showed the superiority of SWI in detecting brain parenchymal TBI lesions, even though the clinical relevance of SWI was debatable. Nevertheless, SWI was still thought to play a meaningful role in predicting a prognosis because of its superior lesion-detection ability. Several studies have shown that the actual lesion site has prognostic value in TBI, especially for lesions in the brainstem, frontal lobe, temporal lobe, basal ganglia, and corpus callosum [781223]. Since the additional lesions detected on SWI in this study were present in these sites, except the corpus callosum, SWI is thought to also have prognostic value and should be considered useful in dogs with TBI.

In summary, this is the first report describing the use of conventional MRI and SWI for the evaluation of TBI lesions in dogs. While the severity of early clinical status was not correlated with SWI results, this type of sequence showed superior ability to detect brain parenchymal TBI lesions in comparison with that of conventional MRI. Because of its characteristics, SWI provides valuable diagnostic information in dogs with TBI, especially in cases that are unexplainable by conventional MRI results. However, conventional MRI also shows superiority for detecting peripheral brain hemorrhage, and its correlation with early clinical signs can help in prognosis prediction. Therefore, examinations that combine conventional MRI and SWI are suggested for dogs with TBI.