Ultrasound, Photoacoustic, and MR Imaging to Study Hyperacute Pathophysiology of Traumatic and Vascular Brain Injury

Abstract

Background and Purpose:

Cerebrovascular dynamics and pathomechanisms that evolve in the minutes and hours following traumatic vascular injury in the brain remain largely unknown. We investigated the pathophysiology evolution in mice within the first three hours after closed-head traumatic brain injury (TBI) and subarachnoid hemorrhage (SAH), two significant traumatic vascular injuries.

Methods:

We took a multi-modal imaging approach using photoacoustic imaging, color Doppler ultrasound, and MRI to track injury outcomes using a variety of metrics.

Results:

brain oxygenation and velocity-weighted volume of blood flow (VVF) values significantly decreased from baseline to fifteen minutes after both TBI and SAH. TBI resulted in 19.2% and 41.0% ipsilateral oxygenation and VVF reductions 15 minutes post injury, while SAH resulted in 43.9% and 85.0% ipsilateral oxygenation and VVF reduction (p<0.001). We found partial recovery of oxygenation from 15 minutes to 3-hours after injury for TBI but not SAH. Hemorrhage, edema, reduced perfusion, and altered diffusivity were evident from MRI scans acquired 90–150 minutes after injury in both injury models although the spatial distribution was mostly focal for TBI and diffuse for SAH.

Conclusions:

The results reveal that the cerebral oxygenation deficits immediately following injuries are reversible for TBI and irreversible for SAH. Our findings can inform future studies on mitigating these early responses to improve long-term recovery.

Introduction

Vascular alterations in the brain can occur over a range of etiologies, from trauma¹ to cerebrovascular disease^2–4 and the resulting damage or rupture of vessels is a major precipitating event, often leading to disability or death.⁵ Although it is known that the earliest outcomes following vascular injury are remarkably consequential, the pathophysiology of this critical time window remains poorly understood. With a deeper knowledge about hemodynamic changes and cellular outcomes during the minutes (min) and hours after vascular damage, diagnostic and therapeutic strategies may be developed to improve long-term outcomes.

As recent evidence suggests, a multi-modal imaging approach can capture the complex vascular and tissue injury coexisting with multiple alterations in physiological metrics in the brain. Differential and quantitative detection of cytotoxic and vasogenic edema, which are common features of brain injury is possible through MRI, and in particular diffusion imaging. Restricted or enhanced water diffusion, normally represented by apparent diffusion coefficient⁶ or diffusion tensor Trace,⁷ can distinguish between these two types of edema. Time-of-flight MRI provides contrast between vessels containing flow of blood and surrounding tissues. Moreover, MRI sequences such as arterial spin labeling (ASL) can reveal altered cerebral blood flow (CBF) following injury, which itself results in compromised oxygen delivery and consumption in tissues.⁸ Moreover, recent advances in photoacoustic imaging⁹ and transcranial Doppler ultrasound¹⁰ enable the quantification of blood oxygen saturation (%sO₂) and hemodynamics, respectively.

Although these imaging modalities could play a significant role in diagnosis and prediction of human outcomes, only sparse imaging data exist within the hyperacute phase after injury.^11,12 Pre-clinical models can provide a more comprehensive picture of the cerebrovascular dysfunction and metabolism alterations in this period. Rodent traumatic brain injury (TBI) studies show focal decreases in CBF within minutes to hours post-injury with recovery and increases seen on later days.^13,14 This reduced CBF is accompanied by increased cerebral metabolism for up to six hours after injury followed by a metabolic depression period for up to 10 days.¹⁵ Recently, A mouse model of TBI showed a reduction in focal %sO₂ derived via photoacoustic imaging six hours following the injury, returning to normal the following day.⁹ A limited number of rodent TBI studies have characterized cerebral edema and perfusion using MRI, starting from 1h post-injury and tracking the injury cascade over days.^14,16 However, to the best of our knowledge, no study has comprehensively explored cerebral hemorrhage, edema, perfusion, and oxygenation in the hyperacute stage following brain injury. Furthermore, the previous studies used direct cortical impact to the brain tissue or similar injury models, which deviates from the vast majority of human non-piercing blunt head trauma.¹⁷

The objective of the current study was to identify the pathophysiology of two brain injury subtypes – subarachnoid hemorrhage (SAH) and moderate TBI – in the hyperacute period (minutes to hours) post injury in mice. We used a combination of ultrasound, color Doppler, photoacoustic, and MRI imaging from baseline up to 3 hours post injury for this purpose. Implementation of SAH via an endovascular perforation model allowed us to isolate the early outcomes of severe vascular injury due to direct vascular rupture and evaluate those against vascular injury induced by a moderate closed-head controlled cortical impact TBI model. By taking a multi-modal imaging approach, we characterize the hyperacute changes in global and local injury markers including blood %sO₂, blood flow velocity, hemorrhage, edema, and CBF.

Methods

Animals

We complied with Animal Research: Reporting of In Vivo Experiments guidelines in conducting the experiments, post-processing and reporting of the results. We purchased Male and female BALB/C mice (n = 24, 12 female, age on day of injury = 15.3 ± 2.6 weeks) from Jackson Labs and Charles Rivers Laboratories and divided them into four groups: TBI, TBI-sham, SAH, and SAH-sham (three males and three females per group). The Institutional Animal Care and Use Committee at the University of Arizona approved all animal procedures and we performed all experiments according to said guidelines and regulations. Because the course of the experiments took several months, we did not order and randomize the animals at once in the beginning. However, we strategically ordered and used animals within a narrow age range over the course of the study and placed each in respective study cohorts on experiment day without any predetermination.

A sample size of 6 subjects per group was planned based on an expected effect size (partial η²) of 0.4 for the primary outcome measure (oxygenation), calculated to yield 80% power to detect significant change in oxygenation over three time points with a p-value <0.05 using analysis of variance (ANOVA): repeated measures, within-between interaction power analysis. Due to the nature of tracking outcomes over the hyperacute window after injury, team members in charge of handling the injuries and imaging were not blinded to the study groups. However, we performed all processing except drawing local pathology regions of interest (ROI) to compute injury extents using algorithms unbiased towards the study groups.

Injury and Scanning Timeline

For injury manipulations, and baseline and post-injury imaging sessions, each animal was deeply anesthetized by isoflurane inhalation. We used a 3% level of isoflurane for induction and kept the level in the 1–2.5% range with the goal of retaining the respiration rate in the 40–60 breaths per minute range. The carrier for the anesthetic was 100% oxygen for injury/sham procedures and 50% oxygen and 50% air for imaging delivered with a total flow rate of 1 L/min.^18,19 We chose a combination of oxygen and air for the prolonged imaging sessions to minimize the compounding hyperoxic effects. We collected MRI baselines on a separate day prior to injury. On the day of injury, we removed the animal’s scalp to minimize signal artifacts for obtaining photoacoustic scans. We collected ultrasound, color Doppler and photoacoustic scans [referred to ultrasound (US)/photoacoustic (PA) in short] immediately prior to injury, then induced the injury (either TBI or SAH) and noted the time (the exact time of impact for TBI, or when SAH microfilament ruptured the vessel near the Circle of Willis). The sham protocols were similar to the injury protocols except for the head impact or vascular perforation. Approximately 15 minutes after the injury induction, we moved the animal back to the ultrasound instrument and collected the 15-minute post-injury US/PA scan. We transferred the animal to the MRI scanner approximately 60 minutes post-injury and allowed the animal’s vitals to stabilize. The MRI scans began 90 minutes post-injury and took approximately 1 hour including setup and pre-scans. The animal transfer between the US/PA and MRI scanners took less than one minute. After the MRI scans ended, we transferred the animal back to the US/PA instrument to acquire the 3-hour post-injury US/PA scans (Figure 1A).

Figure 1.

Overview of the experimental design and utilized imaging modalities: MRI and ultrasound (US). A) Timeline of experiments and imaging sessions. B) Schematics of closed head traumatic brain injury (TBI) and subarachnoid hemorrhage (SAH). C, D) Breakdown of multi-modal imaging approach.​ Short forms “m” and “h” denote minutes and hours, respectively.

TBI Model

To induce moderate TBI, we utilized a previously established closed-head controlled cortical impact model in mice (Figure 1B).^20,21 We installed an impactor (Leica ImpactOne, Leica Biosystems, IL) onto a stereotactic frame. To ensure equivalent lateral impact to each mouse, we adjusted the impactor to an angle of 10 degrees tilted from the vertical direction. We then placed the animal inside the stereotactic frame and delivered anesthesia using a cone-shaped tube. To ensure head support without removing all degrees of freedom and elevate the skull to be level with the flat tip of the impactor, we positioned a 5-mm thick stack of folded paper towels underneath the mouse’s head. We delivered the TBI injury 2 mm to the left of bregma using a 5-mm-diameter impactor tip, terminal velocity of 6 m/s, depth of 2 mm, and dwell time of 100 ms, based on previously published values.^20,21 In the sham protocol, we kept the mice under anesthesia and placed them under the impactor for the same duration as the injury protocol but did not apply any impact. We verified location of impact and head movement trajectory by high-speed videography (Chronos 1.4, Kron Tech., Canada) at a temporal resolution of 10,000 frames/s.

SAH Model

We performed subarachnoid hemorrhage on the mouse brain using a previously established endovascular perforation model (Figure 1B).²² The surgery involved placing the mouse supine and making an initial midline incision along the neck. After incision, we dissected and permanently ligated the left common and external carotid arteries (CCA and ECA, respectively); and then closed the internal carotid artery (ICA) with a temporary suture. Following temporary closure of the ICA, we made a small incision along the CCA, between the CCA ligation point and the branching point of the ECA and ICA. We inserted a 5–0 nylon suture (Ethilon, Ethicon Inc., USA) with a non-coated blunt end through the CCA incision and advanced the filament through the reopened ICA into the brain until we observed resistance. After reaching a resistance point, we pushed the filament further into the brain which induced endovascular perforation of the Circle of Willis near the middle cerebral artery (MCA),²² at which point we retracted and removed the filament from the artery and permanently ligated the CCA posterior to the incision for filament insertion. We performed SAH-sham surgeries by dissecting the CCA and its branches, permanently ligating the CCA and ECA, and temporarily suturing the ICA, which we then removed at the end of surgery. We did not use intracranial pressure monitoring due to its invasive nature. Nonetheless, other studies have forgone the use of intracranial pressure monitoring during SAH surgery and have still reliably induced SAH.^23,24 We verified the existence or absence of a local hemorrhage in the SAH or SAH-sham, respectively, by inspecting the MRI T2*-weighted images taken at the 90-minute post-injury time point, as described below. The criterion for exclusion was absence of hemorrhage for SAH or existence of hemorrhage in SAH-sham.

Multi-modal Imaging and Analysis

Ultrasound and Photoacoustic Imaging

We evaluated in vivo longitudinal variations in cerebrovascular oxygenation (%sO₂), blood flow velocity, and blood perfusion in injury and sham protocols using a battery of 3D ultrasound, color Doppler and photoacoustic scans (Figure 1C). All US/PA protocols were performed using a Vevo 3100 LAZR-X system (FUJIFILM VisualSonics, Toronto, Canada) with a 21 MHz frequency transducer (Vevo MX 250) and the 680–970 nm laser port.

MRI

We acquired an MRI battery of scans to measure hemodynamic and physiological outcomes at baseline and after injury (Figure 1D). We performed all mouse brain imaging using a pre-clinical 7T MRI system (Bruker, Billerica, MA) with mouse-specific head receiver coil and ancillary equipment. We used a multi-echo rapid imaging with refocused echoes sequence for T2-mapping and a multi-gradient echo sequence to collect T2*-weighted images. Time of flight (TOF) angiography images were acquired using a 3D gradient echo pulse. Our diffusion tensor imaging approach used a single shot echo planar imaging pulse sequence. Finally, we collected three axial 2D CBF maps using a pseudo-continuous ASL sequence at 3 landmarks: two 1-mm slices at the bregma and one at the posterior cerebral arteries.

Image Analysis and Post Processing

We quantified %sO_2, Trace of the diffusion tensor (measure of diffusivity) and CBF on ipsilateral and contralateral sides of injury. We drew ROI masks to include the 2–3 mm cortical region for maps of %sO₂ and Trace and separated these masks to ipsilateral and contralateral regions at the midsagittal plane. Next, we computed the average %sO₂ and Trace in each region by multiplying the mean value by the area of the 2D mask in an axial slice, summing over the slices across the brain, and dividing the resulting value by the combined area of the included regions. We used a similar approach to analyze CBF maps by summing over the three collected axial 2D images.

To analyze the color Doppler data, we divided the 3D color Doppler images into two hemispheres, using the underlaid B-mode image as the anatomical reference. In each hemisphere, we multiplied the absolute value of each voxel’s measured velocity by voxel volume to obtain the velocity-weighted volume of flow (VVF) metric.

We also created 3D ROI for pathology in T2 and Trace maps for SAH and TBI using ITK-SNAP software²⁵ by locating abnormal signal within the brains and quantifying its volume.

Statistical Analysis

For statistical analysis, we performed mixed repeated-measure ANOVA with the Greenhouse–Geisser correction to account for violations of sphericity on %sO₂, VVF, CBF, and Trace values, using MATLAB. The within-subject factor was time, and the between-subject factor was injury category (interaction of TBI/TBI-sham and SAH/SAH-sham with time). Given a Greenhouse-Geisser p value smaller than 0.05 for the time intercept and injured/control:time interaction, we performed post-hoc Tukey-Kramer tests for multiple comparisons across time and between corresponding injury/control groups, respectively. We used partial η² and Hedges’ g (with correction for the small sample size) as measures of effect size for the mixed repeated-measure ANOVA and multiple comparisons, respectively. While we included an equal number of both males and females in each cohort in our study design to account for sex-dependent variations, we did not investigate sex as a covariate in our analysis due to the small sample size. Differences with p values smaller than 0.05 were deemed statistically significant.

Results

As planned in the study design and outlined in the Methods section, we used hypointense T2*-weighted signal from MRI scans as a marker for hemorrhage and excluded two animals from the SAH group due to lack of hemorrhage and one animal from the SAH-sham group due to presence of hemorrhage, likely from unintended vascular perforation.

Qualitative oxygenation and blood flow alterations after TBI and SAH

Both TBI and SAH led to conspicuous reductions in %sO₂ and blood velocity that were discernable by eye on oxygenation and color Doppler maps. The TBI-sham group did not show any abnormal pattern at any time point (Figure 2A), whereas the TBI group showed a substantial reduction in %sO₂ at the site of impact and across the brain 15 minutes post-injury, partially recovering at the 3-hour time point (Figure 2B). Animals that underwent arterial occlusion but not perforation (i.e., SAH-sham group) showed reduced ipsilateral %sO₂ and reduced blood flow velocities on the ipsilateral side for both post-injury scans (Figure 2C). However, the outcomes were less severe than for the animal with vascular perforation (i.e., SAH group), which underwent a severe ipsilateral drop in %sO₂ with only a small change on the contralateral side at 15 minutes post injury that progressively worsened at 3 hours post injury. There was a visible reduction in velocity captured from color Doppler imaging in the blood flow velocities of the injured animals for both 15 minutes and 3 hours post injury (Figure 2D).

Figure 2.

Cerebral blood oxygenation (%sO₂) and color Doppler results for traumatic brain injury (TBI) and subarachnoid hemorrhage (SAH) groups as well as shams (A-D). A) TBI-sham does not have any visible changes in either scan. These images qualitatively reveal the hyperacute changes in oxygenation and hemodynamics following the traumatic and vascular injuries. The yellow circle on the oxygenation maps shows the outline of the brain on that slice. Short forms “m” and “h” denote minutes and hours, respectively.

Cerebral blood oxygenation decreases diffusely following TBI and SAH

Brain oxygenation (%sO₂) significantly decreased within whole-hemisphere ROI on the side of injury (ipsilateral) 15 minutes after TBI (19.2% reduction, p<0.001, g=7.18), SAH (43.9% reduction, p<0.001, g=8.05), and SAH-sham (15.5% reduction, p<0.001, g=3.83) compared to baseline (Figure 3A, C). Oxygenation remained significantly lower than baseline for these groups 3 hours after injury as well. In the TBI group, the mice showed some recovery of %sO2 from the 15-minute to the 3-hour post-injury scan (9.3% increase, p=0.004, g=2.26), although the 3-hour post injury ipsilateral %sO₂ was still 11.7% lower than baseline (p<0.001). Compared to the TBI-sham group, ipsilateral %sO₂ values for the TBI group were reduced by 20.0% (p<0.001, g=8.08) and 6.7% (p=0.012, g=1.49) at 15 minutes and 3 hours post injury, respectively. In the SAH group, there was no statistically significant recovery in %sO₂ between 15 minutes and 3 hours post injury on either side of the brain. Compared to the SAH-sham group, SAH ipsilateral %sO₂ values were 34.9% (p<0.001, g=5.19) and 25.6% (p=0.019, g=1.59) lower at 15 minutes and 3 hours post injury, respectively.

Figure 3.

A, C, E, G) Mean oxygenation averaged over the brain on ipsilateral and contralateral sides (* p < 0.05, ** p < 0.01, *** p < 0.001). B, D, F, H) Ipsilateral and contralateral oxygenation per axial slices across 10.5 mm of brain starting from 3 mm anterior to bregma to the posterior part of the brain (solid line: mean oxygenation at slice, shade: standard deviation). Results show hyperacute oxygenation deficits in both injury models, with traumatic brain injury (TBI) showing global changes followed by recovery after three hours and subarachnoid hemorrhage (SAH) showing lateralized effects without recovery. Short forms “m” and “h” denote minutes and hours, respectively.

In addition to the %sO₂ values averaged over the whole-hemisphere ROI, ipsilateral %sO₂ values over 2D axial slices across the brain at different time points also showed unique trends in TBI and SAH injuries. An extended part of mid-brain had reduced %sO₂ 15 minutes post injury in TBI (Figure 3B), while oxygenation values were close to sham levels in the anterior and posterior regions at this time point. On the other hand, the SAH %sO₂ pattern indicated a large deviation from sham levels in the posterior regions for both 15 minutes and 3 hours post injury scans (Figure 3D). Moreover, we did not observe any recovery in either part of the brain by the 3-hour time point.

The contralateral %sO₂ for the TBI group showed a reduction of 17.6% (p<0.001, g=2.06) 15 minutes post injury, which was also 17.68% (p=0.002, g=2.08) lower than TBI-sham (Figure 3E). Contralateral %sO₂ subsequently recovered from 15 minutes to 3 hours post-injury (11.0% increase, p=0.0052, g=0.93) but was still 8.5% lower than baseline (p=0.0107, g=1.48). However, there was no significant difference between TBI and TBI-sham at the 3-hour time point. Contralateral %sO₂ for the SAH group decreased by 11.6% (p<0.001, g=5.15) 15 minutes post-injury and was 21.2% lower than baseline (p=0.008, g=2.00) at 3 hours post-injury (Figure 3G). However, there was no significant difference in contralateral %sO₂ between SAH and SAH-sham at any time point.

Reviewing oxygenation across axial 2D slices for the contralateral side also revealed unique trends for the injury types. There was a similar post-injury reduction in %sO₂ at 15 minutes in the mid brain region for TBI (Figure 3F). Anterior and posterior levels were close to sham levels at this time point. For SAH, however, contralateral %sO₂ levels were in the same range as SAH-sham 15 minutes post injury. These levels deviated from each other particularly in mid brain slices 3 hours post-injury (Figure 3H).

Velocity-weighted volume of flow decreases following TBI and SAH

There was a time-dependent response for all groups in the VVF measure, which we derived by summation of velocity multiplied by voxel volume from the color Doppler image (Figure 4). The TBI group had significant decreases compared to baseline bilaterally for both 15 minutes and 3 hours post-injury scans (Figure 4A). Ipsilateral VVF for TBI was 41.0% (p<0.001, g=2.45) and 45.9% (p<0.001, g=1.95) lower than baseline at 15-minute and 3-hour post injury time points, respectively. The contralateral reductions for TBI with respect to baseline at the same time points were 47.5% (p<0.001) and 40.8% (p<0.001). TBI-shams had significantly different values only between baseline and 3 hours on the contralateral side (37% reduction, p=0.002, g=1.10). The only significant difference between TBI and TBI-sham occurred at the 15 minutes post-injury time point in the contralateral side (54.6%, p=0.039, g=1.15).

Figure 4.

Trends of velocity-weighted volume of flow (VVF) from color Doppler images for A) traumatic brain injury (TBI) vs. TBI-sham and B) subarachnoid hemorrhage (SAH) vs. SAH-sham (* p < .05, ** p < .01, *** p < .001). The results show a decreasing trend from baseline to 3 hours post injury for all study groups, but the SAH and SAH-sham groups both have the most severe ipsilateral drops in VVF. Short forms “m” and “h” denote minutes and hours, respectively.

For the SAH and SAH-sham groups (Figure 4B), on the ipsilateral side, VVF significantly decreased at all post injury time points compared to baseline. Ipsilateral VVF for SAH was 85.0% lower than baseline at 15-minute (p<0.001, g=7.31) and 93.3% lower than baseline at 3-hour (p<0.001, g=12.6) post injury. Ipsilateral VVF for SAH-sham decreased by 55.7% (p<0.001, g=2.72) from baseline to 15 minutes post injury and again by 55.2% (p=0.004, g=1.87) from 15 minutes to 3 hours post injury. Contralateral VVF for SAH was 52.2% (p=0.045, g = 4.03) and 71.1% (p=0.025, g=2.48) lower than baseline at 15-minute and 3-hour post injury time points, respectively.

Three hours after SAH-sham, contralateral VVF reduced by 47.7% (p=0.039, g=1.23) and 44.3% (p=0.002, g=0.81) compared to baseline and 15-minute post injury time points. There were no significant differences in VVF when comparing SAH and SAH-sham.

MRI markers show qualitative focal and diffuse hemorrhage, edema, and ischemia after TBI and SAH

A summary of visual observations in MRI results can be found in Table 1. The baseline scans did not show any abnormalities.

Table 1.

Summary of observations from MRI

Metric/Image	Abnormality	Number of animals with observed abnormality / group size
		TBI-sham	TBI	SAH-sham	SAH
T2	Hypointensity:	0/6	5/6	0/5	1/4
	Hyperintensity:	0/6	6/6	0/5	4/4
T2* weighted	Hypointensity:	0/6	5/6	0/5	4/4
	Hyperintensity:	0/6	1/6	0/5	0/4
Trace (DTI)	Hypointensity:	0/6	5/6	0/5	4/4
	Hyperintensity:	0/6	6/6	0/5	1/4
CBF (ASL)	Ipsilateral/bilateral/focal reduction	0/6	6/6	5/5	4/4
TOF	Reduced ipsilateral vessel intensity (vs. contralateral)	0/6	0/6	5/5	4/4

TBI: Traumatic brain injury, SAH: Subarachnoid hemorrhage, DTI: Diffusion tensor imaging, CBF: Cerebral blood flow, ASL: Arterial spin labeling, TOF: Time of flight.

Following TBI, we generally observed local T2*-weighted (hemorrhage), T2, Trace, and CBF abnormalities at the post-injury MRI scans. The TBI-sham group did not have any signs of pathology in any of the scans (Figure 5A). On the other hand, in a majority (5/6) of the TBI mice, we observed a focal region with low T2 and Trace values and T2*-weighted hypointensities in a core region near the impact site surrounded by small adjacent regions of increased T2 and Trace (Figure 5B). ASL maps showed reduced CBF near the controlled cortical impact (CCI) site (4/6) or diffusely (2/6) following TBI while TOF maps did not indicate any visible change in vessel intensity in the major arterial vessels. Additional abnormalities appearing less frequently included focal increase in T2, T2*-weighted, and Trace (1/6, Figure 6A), abnormalities distant from the impact site in the ventral part of the brain near the olfactory bulb (2/6, Figure 6B) and contusions surrounded by vasogenic peripheries at several different cortical locations (1/6, Figure 6C).

Figure 5.

Images detailing the MRI outcomes of traumatic brain injury (TBI) and subarachnoid hemorrhage (SAH) and sham cohorts (A-D). The images include T2 relaxation, T2*-weighted (T2*w), diffusion tensor Trace, cerebral blood flow (CBF), and time-of-flight (TOF) angiography, all acquired at baseline and 90 minutes following the injury or sham procedure. The focal and diffuse effects from the two injury models are clear in the variety of images. The white arrows point to focal pathologies. Short forms “min” and “ms” denote minute and millisecond, respectively.

Figure 6.

Spatially distinct MRI and oxygenation (%sO₂) findings for traumatic brain injury (A-C) and subarachnoid hemorrhage (D-F). CBF denotes cerebral blood flow. w: weighted; m: minute; h: hour

The SAH-sham group did not have any T2, T2*-weighted or Trace abnormalities (Figure 5C) but did exhibit ipsilateral CBF reductions and reduced signal intensity in the CCA on TOF images suggesting successful occlusion in the SAH-sham group. All the remaining animals in the SAH group (4/4) exhibited hypointense T2*-weighted signal within the brain (Figure 5D and 6D–F) and varying extents of increased T2 and reduced Trace values. ASL maps showed reduced CBF for all animals (4/4), extending bilaterally in three of four animals. TOF scans for all four SAH cases visibly showed reduced vessel intensity in the ipsilateral CCA.

Abnormal T2 and Trace regions are more focal for TBI than SAH

Visually, SAH appeared to affect a more extensive region of the brain, while MRI abnormalities after TBI were focal. To assess this observation quantitatively, we calculated the volumetric extent of abnormal T2 and Trace values for the two injury types (Figure 7A). The extent of regions with increased T2 and decreased Trace following SAH were 16.58±24.43 mm³ and 13.44±12.35 mm³, respectively. Following TBI, increased T2 was observed with an extent of 2.23±1.23 mm³ while increased Trace had an extent of 1.74±1.57 mm³.

Figure 7.

A) Extents of abnormal T2 and Trace regions in traumatic brain injury (TBI) and subarachnoid hemorrhage (SAH) groups. B, C) Cerebral blood flow (CBF) averaged across three 1-mm slices for TBI vs. TBI-sham, and SAH vs. SAH-sham. D, E) Diffusion tensor Trace averaged over the cortical regions on ipsilateral and contralateral sides for TBI vs. TBI-sham, and SAH vs. SAH-sham (* p < 0.05, ** p < 0.01, *** p < 0.001). The short form “m” denotes minutes.

CBF values were bilaterally reduced following SAH but not TBI

TBI and TBI-sham did not show any significant changes between or within groups for hemispheric ROI values of CBF measured by ASL (Figure 7B). On the other hand, there were bilateral decreases in CBF averaged over the three ASL ipsilateral and contralateral slices post-injury in the SAH group compared to baseline (Figure 7C). Post-injury ipsilateral and contralateral CBF values for SAH were 71.4% (p<0.001, g=2.98) and 52.2% (p=0.009, g=1.65) lower than baseline. The SAH-sham group also had a 45.9% decrease (p=0.005, g=3.38) on the ipsilateral side compared to baseline. The ipsilateral CBF in the SAH group was 34.8% lower than SAH-sham post injury (p=0.036, g=1.36).

Hemispheric diffusivity decreased following SAH but not TBI

There was no significant difference in Trace for hemispheric ROI values between TBI and TBI-sham, nor was there any difference for either group between baseline and post-injury values (Figure 7D). Conversely, post-injury ipsilateral and contralateral Trace were 17.2% (p=0.001, g=1.42) and 6.9% (p=0.003, g=1.51) lower than baseline for the SAH group (Figure 7E). Comparing SAH and SAH-sham post-injury values, the SAH group had significantly lower Trace compared to SAH-sham on the contralateral side (2.9% reduction, p=0.015, g=1.68). However, we did not observe any significant difference between SAH and SAH-sham for the ipsilateral side post injury.

Discussion

We characterized the hyperacute vascular and tissue responses to two related experimental injuries with cerebrovascular involvement (TBI and SAH), as early as 15 minutes up to 3 hours post-injury using a combination of color Doppler, photoacoustics, and MRI. The primary photoacoustic findings for TBI were extensive and bilateral decreased %sO₂ at 15 minutes post injury with partial recovery 3 hours post injury. The main MRI outcomes after TBI indicated focal T2, T2*-weighted, and Trace abnormalities, possibly due to hemorrhage and edema pathology and more extensive perfusion deficit post injury. In the case of SAH, cerebral %sO₂ was reduced bilaterally following SAH, but more severely on the ipsilateral side 15 minutes post injury and, unlike TBI, %sO₂ remained significantly lower than baseline even 3 hours post injury. Abnormal biomarkers from MRI outcomes were more prominent and extensive including focal hemorrhage with varying extents of edema and diffuse ipsilateral perfusion deficit.

Hemorrhage and Edema

Focal hemorrhage and the consequent degradation of blood products are hallmarks of physical injury to the tissue. Deoxyhemoglobin is strongly paramagnetic due to its four unpaired electrons and thus has a strong T2*-weighted relaxation shortening effect.²⁶ Therefore, hemorrhage and onset of hematoma are associated with T2*-weighted signal loss and hypointense regions in the T2*-weighted scans. Human MRI observations in the acute phase support the existence of a hemorrhagic core in contusion injuries⁶ while acute clinical^27,28 and hyperacute preclinical studies^23,29 report hemorrhage and hematoma outcomes in SAH. The T2*-weighted images in our study revealed focal hemorrhage in both TBI and SAH models. The majority of the TBIs had a focal hemorrhagic core at the site of impact in the cortex while the focal hemorrhage for the SAH model was deeper inside the brain and was likely due to blood flooding into perivascular spaces and pooling into the parenchyma.

The edema pathology coexisting with this hemorrhagic injury was also uniquely different between the TBI and SAH models. MRI observations rely on diffusivity changes as biomarkers for edema. In this context, enhanced and restricted diffusion can be interpreted as vasogenic and cytotoxic edema, respectively. A clinical study revealed the formation of a vasogenic edema rim surrounding the contusion core, although the earliest scans belonged to 50 hours post-injury.⁶ Mouse and rat studies of open skull CCI have reported vasogenic edema formation as early as 1h post injury supported by T2 and ADC maps but without a hypointense T2/ADC core.^14,30,31 Our current results from a closed-head moderate TBI also show the formation of vasogenic edema around a hemorrhagic contusion core for the TBI model.

On the other hand, the edema resulting from SAH was cytotoxic. Clinical studies have shown the existence of cytotoxic edema in the acute phase following SAH,^26,27 reporting that vasogenic edema is a later outcome of it.²⁶ Our SAH results demonstrate cytotoxic edema pathology in the hyperacute stage after vascular rupture. As opposed to TBI, edema outcome after SAH was not limited to the area immediately surrounding the hemorrhage and was spatially variable.

Perfusion, Hemodynamics, and Oxygenation

Reduced cerebral perfusion could be due to either local tissue damage and subsequent microvascular dysfunction or a decrease in global blood supply to the brain. We observed focal reductions in tissue perfusion in our TBI model. These low-perfusion regions were larger than the hemorrhagic core of impact and the surrounding edema rim. A recently proposed idea suggests an early outcome of TBI is the formation of a traumatic low-perfusion penumbra around the contusion core in focal lesions.³² The kinetic energy absorbed during the impact excessively deforms the microvessels, resulting in the contusion core, while the surrounding penumbra absorbs energy to a smaller extent³³ but still enough to compromise blood brain barrier permeability and lead to vasogenic edema.³⁴ Previous human and animal studies have shown that this expanded area has reduced perfusion in the acute phase and its size correlates with the extent of necrotic tissue determined at later follow-up observations.^35,36 This hyperacute pathological mechanism and its relationship with possible neuronal changes and neurovascular dysregulation require further research.

Injury also altered blood flow velocities in the brain. One of the most common non-invasive methods to measure local velocity in the brain is transcranial ultrasound. In a clinical study, low velocities in the MCA, acquired upon admission of mild to moderate TBI patients, were associated with later neurologic deterioration.³⁷ On the other hand, local MCA blood velocity increases measured using transcranial ultrasound occur with vasospasm following SAH.^38,39 However, a global measure of blood flow velocity in the brain has not been used in the literature. We used color Doppler to measure blood flow velocity in the major vessels and extracted VVF values. These results showed longitudinal reductions for both TBI and SAH, with minimal velocity signal on the ipsilateral side for SAH. The ipsilateral VVF for both SAH and SAH-sham showed reductions compared to baseline, which was expected given the permanent ligation of ECA and CCA for both groups, as well as the hemorrhagic injury in the SAH group.

Our results revealed the diffuse reduction in tissue level perfusion as a hyperacute outcome of SAH. Considering perfusion and VVF results together allows us to understand the relationship between large vessel hemodynamics and tissue level perfusion. Baseline vs. post-injury comparisons for both CBF and VVF showed significant reductions after the injury for SAH and SAH-sham. However, while both SAH and SAH-sham had VVF reductions, their post-injury VVFs were not statistically different from each other, and only the ipsilateral perfusion in SAH was lower than SAH-sham. In other words, the surgery and sham operations both affected major vessel hemodynamics. However, the hemorrhagic event resulted in significantly lower tissue-level perfusion in SAH compared to SAH-sham. A plausible explanation for this trend is the eventual global increase in intracranial pressure due to SAH. Once the subarachnoid space and subsequently the ventricles are filled up by blood on the ipsilateral side, there will be extravasation to the contralateral hemisphere after a few hours. It is at this time point that we see a global perfusion deficit for SAH, which involves hemorrhage, and not SAH-sham, which only involves CCA ligation. Moreover, the hyperacute response after cerebral hemorrhage has been more associated with microcirculatory dysfunction in the form of microthrombosis and blood-brain barrier damage, whereas large artery spasms are commonly observed days after the initial vascular rupture.⁴⁰

Decrease in %sO₂ after TBI is a common secondary injury. It may indicate a delay in energy-dependent recovery due to unbalanced oxygen consumption and delivery temporarily disrupted by the mechanical injury.⁴¹ Clinical studies show that TBI patients with systemic hypoxia prior to or on admission to hospital have higher mortality and poorer outcomes than patients with normal %sO₂ levels.^42,43 Our study is the first to quantify the reduction in arterial blood %sO₂ in the brain as early as 15 minutes after the insult.

Review of the literature reveals possible explanations for the immediate global decrease and subsequent recovery of cerebral %sO₂ in our moderate TBI model. The immediate response to TBI is associated with reduced tissue-level perfusion^13,14 and increased cerebral metabolism of glucose.^15,44,45 Both events could explain the global decrease in %sO₂ 15 minutes post-TBI in our study. Cerebral metabolism of glucose recovers 1–6 hours post injury and goes into a depression period afterwards that could take up to 10 days.¹⁵ The early turning point in metabolism trend may explain the normal global values seen in our 3 hours post injury final %sO₂ scans. It is worth noting that another study⁹ tracked focal %sO₂ values after TBI at baseline and 6 hours post injury and showed statistically significant difference between these two times. They saw recovery to baseline values on day 1 post injury. In our study, The TBI group still had significantly reduced ipsilateral %sO₂ compared to both its own baseline and TBI-sham at 3 hours post injury. However, their injury model was juvenile closed-head mild TBI and they computed %sO₂ only at the left somatosensory cortex, where the contusion happened, whereas we used adult mice and averaged %sO₂ over the brain hemisphere.

The SAH %sO₂ decrease in our study was highly lateralized and persisted throughout the scanning timeline of our experiment. The SAH post-injury ipsilateral %sO₂ values were smaller than SAH-sham, demonstrating the effect of vascular rupture, hemorrhage, and the likely reduction in perfusion pressure against mere ligation of supplying vessels in the sham procedure. Lack of significant contralateral difference between the SAH and SAH-sham %sO₂ responses over time indicates the highly lateralized outcome of this type of hemorrhagic injury.

Limitations and Future Work

Our study had several limitations. (1) The animals had to remain under anesthesia for at least four hours on the injury day. Prolonged anesthesia potentially affects some of the physiological metrics that we tracked, as evident in the %sO₂ and VVF trends for the TBI-sham group. Fluorinated anesthetic agents such as isoflurane affect hemodynamics and primarily elevate cerebral perfusion levels through their vasodilatory impact.^46–48 Research has shown that keeping the induction and maintenance isoflurane levels as low as possible reduces the vasodilatory effects.⁴⁸ Most other anesthetics used for rodent imaging also affect hemodynamic regulation and the ones that do not either have toxicity implications or are not suitable for prolonged sessions.⁴⁸ In our study, the anesthesia period we chose was the shortest possible duration to do the injury and the comprehensive hyperacute imaging until three hours post-injury. We initially experimented with induction and maintenance levels for the mouse strain and injuries covered in this work and used a 50/50 combination of oxygen and air as the carrier gas to minimize the vasodilatory effects of oxygen. This strategy allowed us to pick the combination that kept the animals, especially the injured ones, stably and sufficiently anesthetized throughout the prolonged imaging sessions. We should also emphasize that our study was non-survival and future work targeting the hyperacute window as well as late timepoints should have a well-planned approach to ensure full recovery of animals from the prolonged anesthesia on injury day. (2) Signal to noise ratio decreases with increasing depth in photoacoustic imaging. That is why we mainly drew our %sO₂ ROIs down to 2 mm below the skull and averaged the values in this ROI. Improvements in photoacoustic imaging could allow resolving deeper parts of the brain in the future.

Future directions following this research could include: (1) acquiring the ultrasound-based scans with shorter intervals in the hyperacute phase post injury, which allows tracking the time course of cerebral %sO₂ and VVF response and identifying key times in evolution/recovery of injury, and (2) developing methods to co-register ultrasound and MRI images, which enables us to perform voxel-wise analysis of outcomes using biomarkers from both modalities.