Abstract Text: Introduction Acute cerebral edema is a pathological brain swelling process that may develop in localized or focal regions after brain parenchymal injuries. While cerebral edema can be obvious on computed tomography (CT) images in moderate to severe TBI cases, brain swelling is difficult to detect in mild TBI patients because the brain parenchymal signal is often normal. In this study, we measured longitudinal changes in ventricle size in TBI patients using magnetic resonance imaging (MRI) to serve as a marker of acute generalized cerebral edema. We hypothesized that acute generalized cerebral edema is relatively common in mild head injury cases (i.e., those with normal CT). We also examined the incidence of generalized cerebral edema across TBI severity and presentation on imaging.
Methods TBI patients (age: 19-90 years) had 3D T1-weighted (T1w) MR brain images acquired at baseline ( < 48hr) and 1-week follow-up time points. Subjects with scans containing severe motion or other imaging artifacts were excluded. In total, 115 subjects were included in the analysis and split into 3 groups based on the presentation of whether radiologic findings were present on CT and MR imaging: 42 were CT and MR negative (CT-/MR-), 39 were CT negative and MR positive (CT-/MR+), and 34 were CT and MR positive (CT+/MR+). Additionally, 22 healthy subjects with repeat T1w MR scans at approximately a 1-week interval were analyzed as a reference group to evaluate the reliability of ventricular volume change measurement. The classification of TBI severity (silent, mild, moderate, and severe) was performed based on DoD/VA TBI guidelines. "Silent" TBI patients referred to head injury concerning enough to obtain a CT but no evidence of post-traumatic amnesia or loss of consciousness. To quantify ventricular volume change, image preprocessing included N4 bias field correction, resampling the acquired T1w images to 1 mm resolution, unbiased rigid registration using FreeSurfer, skull stripping using ROBEX, and segmentation using MALPEM. The nonlinear registration was performed using the SyN algorithm in ANTs to obtain a deformation field between two time points. Volumetric change ("dV" where dV=1 represents no volumetric change) was computed by taking the determinant of the gradient of the deformation field within the ventricle label.
Results Among CT- patients, 42 subjects were MR- (silent: 9; mild: 29; moderate: 2; severe: 2) while 39 subjects were MR+ (silent: 9; mild: 26; moderate: 3; severe: 1). In the TBI cohort, the dV was significantly higher in the moderate group than in the silent and mild groups (p < 0.05). The dV for either MR+ or MR- subjects in CT- patients was significantly greater than 1 based on signed rank tests (p < 0.05), indicating increases in ventricular volume between baseline and follow-up. The dV for MR+ in CT- patients tended to be higher than the dV for MR- in CT- patients, but they were not significantly different (dV=1.0171±0.0317 and 1.0110±0.0272, correspondingly; p=0.41). The mean dV during scan-rescan in the reference data set was 1.0003±0.0149. A threshold of the dV around the mean for the controls was set as 2 times of the standard deviation to classify subjects as exhibiting generalized cerebral edema. Among CT- patients, 19.0% of MR- subjects had generalized cerebral edema (8/42) while 28.2% of MR+ subjects had generalized cerebral edema (11/39).
Discussion Ventricles are an intracranial structure that is easily identified on MRI, and its change can be straightforwardly quantified. Because ventricles are much smaller than the brain, small changes in brain size from edema result in relatively larger ventricular changes. These changes were observed in a substantial number of acute TBI patients, even those with MR-. Our results showed that some CT- patients could have dV significantly greater than 1. This significant increase in ventricular volume was seen over approximately 1 week, suggesting that generalized cerebral edema is present acutely and resolves. Therefore, dV could potentially serve as a biomarker for acute generalized cerebral edema. One limitation is that if the edema has not fully resolved within one week, we may underestimate the magnitude of the edema. Additional longitudinal scans of TBI patients after follow-up examination of clinical symptoms merit further investigation.
Keywords: Ventricular Volume, Magnetic Resonance Imaging, Acute Cerebral Edema, Nonlinear Registration, Deformation Field