Hypothermia selectively protects the anterior forebrain mesocircuit during global cerebral ischemia.

Hypothermia is an important protective strategy against global cerebral ischemia following cardiac arrest. However, the mechanisms of hypothermia underlying the changes in different regions and connections of the brain have not been fully elucidated. This study aims to identify the metabolic nodes and connection integrity of specific brain regions in rats with global cerebral ischemia that are most affected by hypothermia treatment. 18F-fluorodeoxyglucose positron emission tomography was used to quantitatively determine glucose metabolism in different brain regions in a rat model of global cerebral ischemia established at 31-33°C. Diffusion tensor imaging was also used to reconstruct and explore the brain connections involved. The results showed that, compared with the model rats established at 37-37.5°C, the rat models of global cerebral ischemia established at 31-33°C had smaller hypometabolic regions in the thalamus and primary sensory areas and sustained no obvious thalamic injury. Hypothermia selectively preserved the integrity of the anterior forebrain mesocircuit, exhibiting protective effects on the brain during the global cerebral ischemia. The study was approved by the Institutional Animal Care and Use Committee at Capital Medical University (approval No. XW-AD318-97-019) on December 15, 2019.

Chinese Library Classification No. R454.5; R743; R318.52

Introduction

Cumulative evidence has shown that hypothermia provides protection of cerebral function, especially in global cerebral ischemia (GI) following cardiac arrest (CA) (Massaro et al., 2013; Randhawa et al., 2015). In the renewed guidelines, therapeutic hypothermia (TH) was the recommended care for post-CA in comatose survivors of CA (Nguyen et al., 2018). The outcomes of CA have improved in the era of TH management (Larribau et al., 2018; Nguyen et al., 2018). Ninety-two percent of patients experience a return to normal or near-normal neurological function (Mooney et al., 2011). Clinical results have provided strong evidence that hypothermia effectively improved the outcomes in patients with CA, stroke or traumatic brain injury (Wu and Grotta, 2013; Kowalik et al., 2014; Wu et al., 2016). Two large randomized controlled trials reported that 32–34°C hypothermia for 24 hours reduced epilepticus seizure (Legriel et al., 2016), and that hypothermia after cardiopulmonary resuscitation resulted in no or minor brain dysfunction (Arrich et al., 2016). On the basis of the importance of hypothermia in GI, previous studies have investigated its underlying cerebral protective mechanisms based on the associated RNA, protein and cytokine levels (Han et al., 2012; Park et al., 2013; Carlin et al., 2017; Wang et al., 2018; Font-Belmonte et al., 2020). Hypothermia also protects the neuronal mitochondria as well as the neurovascular unit in GI (Liu et al., 2016; Carlin et al., 2017; Tang et al., 2020). In a prospective multi-center large cohort study on hypothermia, proton magnetic resonance imaging (MRI) of the thalamus (Ts) showed a clear reduction in incidence of natal encephalopathy (Lally et al., 2019). Three large hypothermia trials have shown that hypothermia reduces both the frequency and severity of brain lesions as seen on MRI (Rutherford et al., 2010; Cheong et al., 2012; Shankaran et al., 2012). However, until now, the protective mechanisms of hypothermia in the functional and metabolic activities of special regions, connection nodes and their associated networks have been poorly understood.

In this study, we used 18F-fluorodeoxyglucose positron emission tomography (18F-FDG-PET) to quantitatively determine the glucose metabolism of different regions in the brain. In addition, we used diffusion tensor imaging (DTI) tractography to reconstruct and explore the cerebral connections protected by hypothermia. Our aim was to investigate the metabolic nodes and connection integrity of a specific circuit in a rat model of hypothermia-induced GI to further elucidate the cerebral protective functions of hypothermia.

Materials and Methods

Animal model

The study protocols were approved by the Institutional Animal Care and Use Committee at Capital Medical University (approval No. XW-AD318-97-019) on December 15, 2019. The animals were purchased from Beijing Vital River Laboratory Animal Technology Co. Ltd., affiliated with Charles River Laboratories (CRL), Beijing, China (license No. SCXK 2001-0017), and were housed at the Animal Care Facilities of Capital Medical University. All experiments conformed to the Guide for the Care and Use of Laboratory Animals (National Institutes of Health Publication 85-23, revised in 1985). All experiments were designed and reported according to the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines. The animals were housed under controlled environmental conditions (12-hour light/dark cycle, ambient temperature 23 ± 2°C and 60–70% humidity).

This study used male animals only because female animals have higher estrogen levels that protect them from infection. Eighteen male Sprague-Dawley rats, specific-pathogen-free level, 9–11 weeks old, weighing 320–350 g, were randomized into three groups. The hypothermia GI group (n = 6) was maintained at 31–33°C to establish the GI model. The temperature probe was inserted into the rectum and the rats were placed on a cooling/warming blanket (CMA 150, Carnegie Medicine, Stockholm, Sweden) that was connected to a thermostatically controlled system (CMA 150, Carnegie Medicine) that maintained the temperature using feedback regulation. During a 25-minute cooling period, alcohol was sprayed on the fur of rats to reduce body temperature. The hypothermia period was maintained for 150 minutes, after which the rats were warmed for 30 minutes, and then maintained at normal temperature until scanned. The normothermia GI group (n = 6) was maintained at 37–37.5°C using cooling/warming blankets with the same system to establish the GI model. Duration of normothermia was maintained for 150 minutes (). The sham group (n = 6) was maintained at 37–37.5°C and subjected to a median incision in the neck but without induction of GI model.

Rectal temperature during the global cerebral ischemia model procedure.

The rectal temperatures in hypothermia and normothermia global cerebral ischemia groups were 32.5 and 37.5°C, respectively. Data are expressed as the mean ± SD (n = 6). Hypothermia group: Hypothermia global cerebral ischemia group; normothermia group: normothermia global cerebral ischemia group. The rats in the hypothermia cerebral ischemia group were moved to initiate rewarming and the temperature probe was removed from rectum. Therefore, the last two temperatures were not recorded.

The 2-vessel occlusion GI model was accomplished using a modification of the Longa's method, as previously described (Smith et al., 1984; Zhang et al., 2013). Catheters were inserted into the external jugular veins to draw blood, the left femoral artery for blood pressure monitoring and the right femoral artery for Heparin (H12020505, Tianjin Biochem Pharmaceutical Co., Ltd., Tianjin, China) infusion. Under an operating microscope (Perlong Medical Equipment Co., Ltd. Nanjing, China), a median incision in the neck was made and the bilateral common carotid arteries were isolated and each proximally encircled by a ligature line. Heparin (150 IU/kg) was administered and blood was drawn via the jugular catheter to decrease the mean arterial pressure to 40–50 mmHg. During the procedure, the vagus nerve was carefully preserved, blood gases were measured and the tidal volume of the respirator during the intubated period was adjusted to achieve an arterial partial pressure of carbon dioxide of 35–45 mmHg, an arterial partial pressure of oxygen > 90 mmHg and a pH value of 7.35–7.45. After approximately 10 minutes of the left common carotid artery ligature, the right common carotid artery was ligated to finally block the cerebral blood flow. After a total of 15 minutes, the blood was slowly re-infused through the jugular catheter, followed by 0.5 mL sodium bicarbonate (0.6 M). After a 135-minute recovery period, the wounds were sutured and the rats were returned to their cages. No animals or data points were excluded from the analysis.

Data acquisition for 18F-FDG-PET, MRI DTI, and T2

The rats were scanned 24 hours after the GI model was established. Before 18F-FDG injection, all rats had access to drinking water at all times but were deprived of food for 12–15 hours. For each rat, 18F-FDG (18.5 MBq/100 g body weight; Atom High Technology (HTA) Co., Ltd., Beijing, China) was administered via tail vein injection without anesthesia. Subsequently, the rats were returned to their cages and kept in a room for 40 minutes with minimal ambient noise for maximization of 18F-FDG uptake in the brain (Caballero Perea et al., 2012; Quinn et al., 2016). Subsequently, the rats were anesthetized using a nose cone with 2% isoflurane in 100% oxygen (IsoFlo, Hebei Jiumu Pharma, Ltd., Langfang, China) for the period of the scan. The rats were placed in the prone position on the scanner bed and with a plastic stereotactic head holder. 18F-FDG-PET images were acquired on an animal PET system (E-plus260, Institute of High Energy Physics, Chinese Academy of Sciences, Beijing, China) at the center of the field of view and a static acquisition of 10 minutes with radial spatial resolution of 1.55 mm full-width half maximum was performed. The images were subsequently reconstructed using the ordered subsets expectation maximization (4 iterations, 12 subsets) algorithm. They were reconstructed on a 90 matrix × 97 matrix × 200 matrix, with a voxel size of 0.5 × 0.5 × 1 mm3. Finally, all scans were saved in the analyze format. DTI data were acquired by a 38-mm birdcage rat brain quadrature resonator for radiofrequency transmission using a 7.0 T animal MRI scanner (70/16 PharmaScan, Bruker Biospin GmbH, Rheinstetten, Germany). DTI images were obtained with 12,000 ms repetition time, 32.248 ms echo time, 163 matrix size, DTI image 128 × 128 × 48 mm3, voxel size 0.35 × 0.35 × 0.56 mm3, with no slice gap. Diffusion weighting was applied along 30 independent axes, with a b value of 1000 s/mm2. Eight reference images with a b value of 0 s/mm2 were acquired. Finally, Paravision 5.0 software programs (Bruker Biospin corporation, Baltimore, MD, USA) converted all original Bruker images to the DICOM format (Li et al., 2016; Zhang et al., 2016).

Analysis of 18F-FDG-PET images

Data analysis of all 18F-FDG-PET images and identification of the significant differences of 18F-FDG signals were performed in a rat statistical parametric mapping (spmrat)-IHEP toolbox (Nie et al., 2013, 2014) in SPM8 (Wellcome, Department of Cognitive Neurology, London, UK). First, using MRIcro manually removed background and the body tissues of all images (Mangin et al., 2016), the origins of the images were repositioned at D3V to correspond to the standard 18F-FDG-PET template (Paxinos and Watson, 2005). By scaling up the voxel size in the Analyze header, the individual rat brain images were normalized into Paxinos & Watson space spatially by a factor of 4 (Casteels et al., 2006; Nie et al., 2010), registering to the 18F-FDG-PET template, the extracranial tissues were removed via the intracranial image and the background cut off by shearing the matrix. Finally, all normalized 18F-FDG-PET images were smoothened by a Gaussian kernel of 2 × 2 × 4 mm3 full-width half maximum. Based on the framework of the general linear model, all smoothened data were voxel-wise analyzed. Based on an unbiased scale factor, proportional scaling and intensity normalization were applied to account for global confounders (Crone et al., 2014). Based on a voxel-level height threshold of P < 0.005, the brain regions with significant 18F-FDG differences from the sham group were yielded. At the cluster level, false discovery rate correction for multiple comparisons was also conducted. The data were analyzed using voxel-based analysis. According to both the sagittal and transverse sections, the realignment of the internal contour of each tract was also translated. The rat brain atlas stereotaxic coordinates were defined such that the x-axis is positive to the right and negative to the left of the midline, the y-axis is positive toward the ventral direction and negative to the dorsal direction, and the z-axis is positive toward the olfactory bulb relative to the bregma and negative in the direction of the cerebellum. The 18F-FDG uptake was compared between the hypothermia/normothermia GI groups and the sham group. If 18F-FDG uptake of the hypothermia/normothermia GI group was lower than sham group, they will have hypo-metabolic regions that will show the cold pseudo-color, blue. If 18F-FDG uptake of the hypothermia/normothermia GI group was higher than sham group, they will have hyper-metabolic regions that will show the warm pseudo-color, orange. If 18F-FDG uptake of the hypothermia/normothermia GI group equals to that of the sham group, the equal-metabolic regions will show the gray pseudo-color, gray. The KE value represents the size of a cluster, specifically the volume of hypometabolic region (Liang et al., 2017).

Analysis of MRI DTI and T2 images

All DTI images were preprocessed in the FMRIB Software Library (FSL) (http://fmrib.ox.ac.uk/fsl). In brief, head motion and eddy currents were removed using FMRIB's Diffusion Toolbox within FSL. The fractional anisotropy (FA) of each individual tract was then calculated. Subsequently, voxel-wise analyses of FA images were performed in a spmrat-IHEP toolbox (SPM8), which is similar to the FDG-PET data analysis (Nie et al., 2013). In brief, the FA data sets were preprocessed by skull stripping and repositioning of the origin point. Then, the FA images were spatially normalized to an FA template image in Paxinos & Watson space (Paxinos and Watson, 2005) and along with MD images. All these FA images were smoothened as described above. Finally, all smoothened data were voxel-wise analyzed based on the general linear model framework. False discovery rate correction for multiple comparisons was also conducted at the cluster level. The higher FA values indirectly reflected conservation of the white fibers.

Statistical analysis

The statistical analysis was performed in spmrat-IHEP (Wellcome Department of Clinical Neurology, London, UK) as follows. Values of the hypothermia or normothermia GI group were averaged and compared with the sham group. The data were analyzed by two-sample t-test. The level of significance was regularly set at P < 0.05. Based on the generalized linear model, the statistical analysis model of the smoothed mean normalized uptake values 18F-FDG-PET data was established and the Student's t-test analysis was performed based on hypothermia GI and normothermia GI groups. The mn-UV was computed in KE, a measure of the size of the cluster as the voxel numbers in the cluster, which directly reflected the volume of the region (Liang et al., 2017). The Tmax value is the maximum t-value in each cluster; a Tmax value > 1 indicates a significant difference. Peak coordinates (mm) are the coordinates of the maximum point in Paxinos & Watson space.

Results

Hypothermia reduces the hypometabolic regions in brain of GI rat

PET results showed the hypometabolic regions of whole brain were significantly smaller in the hypothermic GI group when compared with those in the normo-thermic GI group (Figure and ). When compared with the normothermic GI group, the hypothermia GI group also had significantly smaller hypometabolic regions in the Ts and prefrontal-cortex (PFC) and primary sensory areas (Figure ). These results indicate that hypothermia selectively protects the anterior forebrain (cortical) and Ts (subcortical part) of the mesocircuit under GI.

The hypometabolic regions in the global cerebral ischemia model rats by PET.

(A, B) The hypometabolic regions were significantly larger in the normothermic global cerebral ischemia group (A) when compared with those in the hypothermic global cerebral ischemia group (B). The cold pseudo-color (blue) indicates the hypometabolic regions, while the warm pseudo-color (orange) indicated the hypermetabolic regions. Red lines indicate the position of slice cuts.

The hypometabolic regions in anterior forebrain and Ts of the mesocircuit in the global cerebral ischemia model rats.

The hypometabolic regions in the Ts, PFC and primary SC are smaller in the hypothermia GI group than that in the normothermic GI group. The cold pseudo-color (blue) indicates the hypometabolic regions, while the warm pseudo-color (orange) indicates the hypermetabolic regions. (A, B) The coronal sections of the normothermia GI group (A) and hypothermia GI groups (B). (C, D) The sagittal sections of the normothermia GI group (C) and hypothermia GI group (D). GI: Global cerebral ischemia; PFC: prefrontal cortex; SC: sensory cortex; Ts: thalamus.

The voxel-wise analysis results are presented as KE values in . As the center of the anterior forebrain thalamic mesocircuit (Schiff, 2008), the voxel number in the Ts hypometabolic regions was significantly lower in the hypothermia GI group compared with that in the normothermia GI group (). The PFC is an important region of the mesocircuit (Schiff, 2008). The KE value in the PFC region was also significantly lower in the hypothermia GI group than that in the normothermia GI group (). The results show that hypothermia significantly preserved anterior forebrain neuronal metabolic activity.

Hypothermia-treated GI rats have no obvious thalamic injury

The results from the MRI T2 scans show obvious injury to the Ts region in the normothermia GI group (). There was almost minimal brain injury in Ts region in the hypothermia GI group ().

Difference of brain injury on MRI T2 images between normothermia GI (A) and hypothermia GI (B) groups.

There was obvious damage (red box) to the thalamus in the normothermia GI group. GI: Global cerebral ischemia; MRI: magnetic resonance imaging. Red box indicates the injury part in MRI image.

Difference of connective fibers between hypothermia GI and normothermia GI groups

The DTI MRI indicated the preservation of the corpus callosum and internal capsule (the Ts-cortical white fibers connections) that was also demonstrated by FA in DTI. The voxel-based analysis demonstrated that hypothermia selectively preserved the anterior forebrain mesocircuit, including the Ts, prefrontal cortex (PFC), and the connections including the corpus callosum and internal capsule ().

Discussion

Our study found that the metabolic activity of the anterior forebrain thalamic mesocircuit and the connective tissue fibers were significantly retained in the hypothermia GI group. GI theoretically produces relatively uniform cellular damage across all brain regions, however, we found that hypothermia produced a graded and differential protection in special neuronal populations and different brain regions in response to global ischemia. Hypothermia significantly preserved anterior forebrain neuronal metabolic activity, as demonstrated by 18F-FDG-PET findings. The MRI T2 results also supported this observation. The preservation of the corpus callosum and internal capsule, the Ts-cortical white fibers connections, was also demonstrated by FA in DTI. The PFC, Ts and associated connections were identified as key components of the anterior forebrain thalamic mesocircuit (Schiff, 2008) and our research identified these areas as those most significantly preserved by hypothermia. That the anterior forebrain mesocircuit was selectively preserved indicates that it plays a pivotal role in hypothermia protection from GI damage.

The anterior forebrain thalamic mesocircuit is the core circuit for maintaining the normal consciousness, arousing and sleep-awake circle (Schiff, 2010; Fridman et al., 2014). This mesocircuit plays a key role in recovery from a coma with potentially retained consciousness and provides the conceptual foundation for the central Ts as the privileged node for arousal neuromodulation (Zhang et al., 2013; Lant et al., 2016). Recent research into the anterior forebrain mesocircuit found that a loss of excitatory output from the central Ts to diffuse cortical areas had a causative role in disorders of consciousness (DOCs) (Lant et al., 2016). In a recent neuroimaging study (Song et al., 2018), selective hypometabolism in the cortex and Ts was reported in acquired DOCs induced by GI injury. Recovery from DOCs has been shown to reverse the modulation of widespread excitation across the anterior forebrain and correlates with the restoration of central thalamic output to the prefrontal cortex (Song et al., 2018). Impairments in the anterior forebrain thalamic mesocircuit appear as DOCs, indicating that this mesocircuit may provide a target for restorative therapies in patients with DOCs (Neske, 2015). Some patients diagnosed as conscious awareness and with cognitive function following severe brain injuries can recover after surprisingly long-time intervals of months, years and even decades (Burruss and Chacko, 1999; Macniven et al., 2003; McMillan and Herbert, 2004; Lammi et al., 2005; Voss et al., 2006). One patient, after remaining in minimally conscious state for 19 years following a severe traumatic brain injury, spontaneously recovered full expressive and receptive language and revealed evidence of ongoing structural rehabilitation (Voss et al., 2006). Forgacs et al. (2016) first described that a patient who underwent prolonged CA and standard TH protocol developed isolation syndrome. This phenomenon was also confirmed in neurological prognostication after CA owing to the recent evolution of clinical practice (Lauritzen et al., 2016). Other recent preclinical studies and pilot clinical studies have reported GI following CA and resuscitation with a standard TH protocol shows isolation syndrome (Owen et al., 2006; Song et al., 2018). The effect of TH on the integrative mesocircuit supports the arousal regulation mechanisms of severe brain injury, even after GI. Structural MRI studies may show structural changes within the brain. Functional, metabolic and structural disconnections within the thalamocortical regions of the default node network lead to unconsciousness and are correlated with the clinical severity. The hypothermia treatment actively modulated the metabolic activity of anterior forebrain mesocircuit.

The anterior forebrain and the Ts are two cardinal structures for the mammalian brain (Granato et al., 1995), and the anterior forebrain mesocircuit is highly conserved in the mammalian brain. Hypothermia selectively preserved the anterior forebrain and the Ts when subjected to limited oxygen and energy conditions during ischemia. The anterior forebrain mesocircuit provides the foundation for the late recovery of function following severe brain injury (Schiff, 2008, 2010). The anterior forebrain mesocircuit may be the important “seed region”. Hypothermia selectively protected this region, enabling future recovery and leading to the total recovery of these patients, even several years later. Three possible fundamental pathophysiological mechanisms are likely to account for our findings. First, the highly conserved anterior forebrain mesocircuit has particular neuron and synaptic connections that are different from other neural circuits. Second, that hypothermia leads to the selective protection of special neuronal populations. Third, the anterior forebrain mesocircuit is indispensable for maintaining any hypothermia protection effects (Monti et al., 2015; Boly et al., 2017). The preserved functional integrity of the anterior forebrain provides a specific protective mechanism and potential target for GI treatment.

Nevertheless, there is a caveat that should be considered. We did not test the mesocircuit of female rats with CA conserved following TH. Based on the results in rodents, our next step will be to perform MRI and PET in patients to confirm that this mesocircuit is also preserved by clinically administered hypothermia.

In conclusion, based on 18F-FDG-PET and DTI-MRI findings, we demonstrated that hypothermia can significantly preserve the integrity of the anterior forebrain-thalamic mesocircuit.

Table 1

The KE value in the rat brain in two groups

Anatomical name	KE	T Max	Stereotaxic coordinates
			x	y	z
Normothermia global cerebral ischemia group
Prefrontal cortex	553**	4.2414	-0.7308	3.2066	3.7221
Thalamus	2391**	6.1129	-0.8908	8.389	-13.7979
Corpus callosum	54	4.0773	-1.6465	2.6963	2.2821
Internal capsule	29	3.9995	3.4353	3.1571	2.0421
Sensory cortex	1098*	4.8197	4.1167	2.4048	1.0821
Motor cortex	263	3.8065	3.3049	1.2724	1.8021
Prelimbic cortex	477	4.25	-0.73	3.22	3.48
Tegmentum of pons	3426	6.7024	-0.6268	8.2371	-13.5579
Tegmentum of midbrain	1165	5.8208	-1.9484	6.9627	-4.9179
Hypothermia global cerebral ischemia group
Prefrontal cortex	71**	6.8144	4.5209	6.976	0.8421
Thalamus	342**	5.8853	-0.7704	7.7274	-12.8379
Corpus callosum	4	3.7947	4.1167	3.573	1.0821
Internal capsule	1	3.9995	3.0772	3.995	-1.0779
Sensory cortex	145*	4.5506	3.4353	3.1571	2.0421
Motor cortex	263	3.8065	3.3049	1.2724	1.8021
Prelimbic cortex	477	12.1443	4.5275	7.4567	0.3621
Tegmentum of pons	3474	6.405	-0.6367	7.8811	-12.8379
Tegmentum of midbrain	1105	5.1026	1.292	6.485	-7.3179

KE: The size of a cluster, in which the number, such as 2391, stands for the voxel numbers in the cluster; Tmax value: the maximum t-value in each cluster. Tmax value > 1 means significant difference. x: x-axis, which is negative to the left of the midline and positive to the right; y: y-axis, which is positive to the ventral direction relative to the dorsal direction; z: z-axis, which is positive in the direction of the olfactory bulb relative to the bregma and negative in the direction of the cerebellum. *P < 0.05, **P < 0.01 (two-sample t-test).

Table 2

FA value in the white matter tracts of rat mesocircuit in two groups

Region	Abbreviation	Functional connection	FA value
			Hypothermia GI group	Normothermia GI group
Internal capsule	IC	Part of the connection between the cortex and thalamus	0.408±0.051*	0.384±0.011
Corpus callosum	CC	Part of the connection between the cortex and thalamus	0.710±0.159**	0.458±0.058
Mammillothalamic tract	MT	Part of the connection between the cortex and thalamus	0.610±0.031	0.670±0.070
Forceps minor of the corpus callosum	FMI	Part of the connection between the cortex and thalamus	0.928±0.071*	0.696±0.121
Forceps major of the corpus callosum	FMJ	Part of the connection between the cortex and thalamus	0.278±0.022	0.191±0.019
Genu of the corpus callosum	GCC	Part of the connection between the cortex and thalamus	0.765±0.147	0.596±0.065
Intermedioventral thalamic commissure	IMVC	Intermedioventral thalamic commissure	0.379±0.062	0.457±0.101
Stria medullaris of the thalamus	SM	Medullaris of the thalamus	0.405±0.209*	0.225±0.064
Longitudinal fasciculus of the pons	LFP	Longitudinal section of the connection between the cortex and thalamus	0.678±0.101	0.528±0.051
Medial forebrain bundle	MFB	Forebrain bundle	0.209±0.013	0.194±0.012
Medial longitudinal fasciculus	MLF	Longitudinal fasciculus	0.894±0.050*	0.661±0.131

Data are expressed as mean ± SD (n = 6). *P < 0.05, **P < 0.01 (two-sample t-test). Cluster number: The number of clusters with consecutive voxels with a significant decrease in FA which is assigned sequentially and artificially. FA: Fractional anisotropy; GI: global ischemia.