Volumetric and shape analyses of subcortical structures in United States service members with mild traumatic brain injury

Abstract

Mild traumatic brain injury (mTBI) is a significant health concern. The majority who sustain mTBI recover, although ~20 % continue to experience symptoms that can interfere with quality of life. Accordingly, there is a critical need to improve diagnosis, prognostic accuracy, and monitoring (recovery trajectory over time) of mTBI. Volumetric magnetic resonance imaging (MRI) has been successfully utilized to examine TBI. One promising improvement over standard volumetric approaches is to analyze high-dimensional shape characteristics of brain structures. In this study, subcortical shape and volume in 76 Service Members with mTBI was compared to 59 Service Members with orthopedic injury (OI) and 17 with post-traumatic stress disorder (PTSD) only. FreeSurfer was used to quantify structures from T1-weighted 3 T MRI data. Radial distance (RD) and Jacobian determinant (JD) were defined vertex-wise on parametric mesh-representations of subcortical structures. Linear regression was used to model associations between morphometry (volume and shape), TBI status, and time since injury (TSI) correcting for age, sex, intracranial volume, and level of education. Volumetric data was not significantly different between the groups. JD was significantly increased in the accumbens and caudate and significantly reduced in the thalamus of mTBI participants. Additional significant associations were noted between RD of the amygdala and TSI. Positive trend-level associations between TSI and the amygdala and accumbens were observed, while a negative association was observed for third ventricle. Our findings may aid in the initial diagnosis of mTBI, provide biological targets for functional examination, and elucidate regions that may continue remodeling after injury.

Keywords: Mild traumatic brain injury; Military; Service members; Shape analyses; Subcortical structures; Volumetric measures; mTBI.

Fig. 1

Illustration of subcortical surfaces reported in this study. Note that all images are in radiological convention throughout this manuscript, that is, the right hemisphere appears on the left side of the image

Fig. 2

t-Value maps of the regression of the Jacobian determinant (JD) on the ortho-TBI contrast from a inferior, b anterior and c left perspectives. Note that all images are in radiological orientation, i.e. left–right flipped. The top row plots the average JD within regions differing significantly between TBI and Orthopedic groups. The second row illustrates FDR-thresholded t value maps in which only highlighted regions are significant. The third row maps Jacobian t values across the entire set of surfaces. The bottom row compliments this by mapping the associated t values for the local thickness (RD) measure to determine whether changes in local surface area result from concavity or convexity. Here, the directions of the JD and RD associations agree within regions of significance indicating that regions in red result from dilation while regions in blue are suggestive of atrophy. All coefficient t values had 125 degrees of freedom

Fig. 3

Percent volume change in TBI participants relative to orthopedic controls by region and hemisphere with bootstrapped 95 % confidence intervals

Fig. 4

t Value maps of a the regression of thickness features (RD) on TSI shown from an anterior perspective showing a significant increase in the thickness of the right anterior amygdala with extended TSI. The top row of maps is FDR-thresholded with the remaining highlighted regions being those that are significantly associated. The second row of maps is unthresholded showing the global distribution of associations. Wherever significant associations are present we plot the least squares fit of the association for the vertices surviving FDR correction. All coefficient t values in a had 60 degrees of freedom. b–d Plot associations between the thickness (b) or surface area (JD measure) with the interaction of TSI and age. b Reveals a significant differential association between the thickness of the right amygdala between older and younger participants with the younger cohort having a more rapid expansion of this region than older participants. Identify regions of the c anterior right amygdala and d right pallidum with significantly differing trajectories of surface area changes. In both cases, the older participants exhibit gains in surface area at extended TSI while younger participants show significant reductions in the surface area of these regions. Associations in b–d had 59 degrees of freedom. Note that these figures are in radiological orientation, left–right flipped

Fig. 5

Association of surface area (JD) with method of injury (MOI). Positive associations of surface area and MOI are shown in warmer colors and are seen to be more spatially concentrated in anterior lateral regions of the subcortical structures. While these effects are slightly below statistical significance, the trend of spatial distribution is suggestive of the nature of a blast injury