Structural Integrity of Medial Temporal Lobes of Patients with Acute Mild Traumatic Brain Injury

Abstract

Post-traumatic amnesia (PTA) is an acute characteristic of traumatic brain injury (TBI) and the duration of PTA is commonly used to estimate the severity of brain injury. In the context of mild traumatic brain injury (MTBI), PTA is an essential part of the routine clinical assessment. Macroscopic lesions in temporal lobes, especially hippocampal regions, are thought to be connected to memory loss. However, conventional neuroimaging has failed to reveal neuropathological correlates of PTA in MTBI. Texture analysis (TA) is an image analysis technique that quantifies the minor MRI signal changes among image pixels and, therefore, the variations in intensity patterns within the image. The objective of this work was to apply the TA technique to MR images of MTBI patients and control subjects, and to assess the microstructural damage in medial temporal lobes of patients with MTBI with definite PTA. TA was performed for fluid-attenuated inversion recovery (FLAIR) images of 50 MTBI patients and 50 age- and gender-matched controls in the regions of the amygdala, hippocampus, and thalamus. It was hypothesized that 1) there would be statistically significant differences in TA parameters between patients with MTBIs and controls, and 2) the duration of PTA would be related to TA parameters in patients with MTBI. No significant textural differences were observed between patients and controls in the regions of interest (p>0.01). No textural features were observed to correlate with the duration of PTA. Subgroup analyses were conducted on patients with PTA of>1 h, (n=33) and compared the four TA parameters to the age- and gender-matched controls (n=33). The findings were similar. This study did not reveal significant textural changes in medial temporal structures that could be related to the duration of PTA.

Introduction

Post-traumatic amnesia (PTA) refers to the period of suppressed full consciousness, impaired anterograde memory, attention deficits, and disorientation caused by mechanical energy to the head from external physical forces.¹ PTA is an established hallmark of traumatic brain injury (TBI).^1
–3 The duration of PTA is used to estimate the severity of brain injury.^4,5 According to the majority of diagnostic criteria for TBI, PTA of<24 h is considered to be a characteristic of mild TBI (MTBI). If this PTA is accompanied by a loss of consciousness (LOC) for<30 min and a Glasgow Coma Scale (GCS) score ≥13 30 min post-injury and later upon presentation, the injury can be designated as an MTBI.⁶ In addition to severity assessment, accurate measurement of the duration of PTA is proposed to be important for two more reasons.⁷ First, it is considered to be one of the best general predictors of long-term outcome following TBI.⁸ However, this does not hold true in MTBI. Second, it can be used to make decisions about discharge.⁹ In previous studies,^10
–12 PTA has not been shown to predict neuroimaging findings (CT or MRI), increased symptomology, or neurocognitive impairment following MTBI. In some studies, however, PTA is related to outcome.^13

–16

The anatomic basis of PTA is founded on animal studies.^17,18 Macroscopic lesions in temporal lobes, especially hippocampal regions, resulted in extensive periods of memory loss independent of lesion etiology. The most informative reports about the functional neuroanatomy of human memory come from cases in which there has been an opportunity to conduct extensive neuropsychological testing as well as postmortem neurohistological analysis.^19
–21 The hippocampus has been shown to play a pivotal role in memory formation, especially regarding acquisition and new memories. Interaction between the hippocampal system and the neocortex is considered necessary for memory storage.²² In MTBI with considerable duration of PTA, it is reasonable to assume that hippocampal integrity is compromised structurally or at least physiologically. Nevertheless, conventional neuroimaging techniques (CT and MRI) do not reveal neuropathological correlates of PTA in MTBI.

In MTBI, structural damage to neural tissue, if present, is not often detectable by conventional MRI. Even with recent improvements in MRI techniques, such as diffusion-weighted MRI, diffusion tensor imaging (DTI), and new MRI sequences, the identification of neuropathology associated with MTBI remains challenging.²³ The analysis of image texture parameters might offer useful information about underlying pathophysiology in the brain tissue. Texture can be regarded as a similarity grouping in an image; it is an image feature that corresponds to both brightness value and pixel locations. However, the human eye is not always able to recognize all textures because of texture scale variability. Computer-based texture feature analysis provides a means for extracting additional information from structural brain MR images that cannot be otherwise detected. Texture analysis (TA) is an image analysis technique that quantifies MRI signal changes between individual pixels of an image and, therefore, the variations in intensity patterns within the image.²⁴ The aim of TA is to calculate mathematical patterns (i.e., texture features) that can be used to characterize the underlying histology.²⁵ MRI signal is usually sensitive to changes in brain tissues caused by neurological diseases, which results in signal fluctuations and, therefore, intensity changes in the image, including many that are imperceptible to the human eye. For these reasons, TA can be considered a potential tool in investigative neuroimaging.^{26

–33}

TA has revealed microstructural alterations in neurodegenerative disorders.^{26

–34} TA has been used to characterize benign and malignant brain tumors, and it can discriminate between tissue and tumor constituents.^24,35,36 In epilepsy studies, Yu et al.³⁴ evaluated texture features extracted from proton-density and T2-weighted images of the hippocampi of patients with unilateral temporal lobe epilepsy and of healthy controls. They were able to correctly identify hippocampal sclerosis with proton density MR images by the means of TA. Using histological evidence of hippocampal sclerosis as the gold standard, Bonilha et al.²⁷ evaluated several textural features extracted from T1-weighted MR images acquired from patients with mesial temporal lobe epilepsy, and they were able to discriminate between sclerotic and healthy hippocampi. Texture features have also been used in lesion characterization, and several studies have been published that differentiate multiple sclerosis (MS) lesions from hyperintensities in normal white matter, and differentiate normal-appearing white matter from normal white matter. TA has been proposed as a method for identifying active MS lesions as well as monitoring the progression of the disease.^31,32,37 TA is also capable of detecting tissue alterations in patients with Machado–Joseph disease in areas already shown to be affected, in histopathological studies.³⁰ In the context of TBI, TA has been used in previous studies by Holli and colleagues.^38,39 In these studies, statistically significant differences in texture between the left and right side in the mesencephalon, in cerebral white matter at the level of corona radiata, and in the corpus callosum, were observed in MTBI patients. In addition, significant correlations between texture parameters and memory composite scores were found in the mesencephalic area and in the genu of the corpus callosum.

The objective of this study was to apply the TA technique to MR images of MTBI patients and control subjects and to assess the possibility of observing microstructural damage in medial temporal lobes of patients with MTBI who have definite PTA. This study had two hypotheses. First, that there would be statistically significant differences in TA parameters between patients with MTBIs and controls. Second, that the duration of PTA would be related to TA parameters in patients with MTBI.

Methods

Study framework and ethics

This work is part of the Tampere Traumatic Head and Brain Injury Study. Subjects were enrolled from the Emergency Department (ED) of the Tampere University Hospital between August 2010 and July 2012. The ED provides health services for a joint municipal authority of 22 municipalities (both urban and rural), with a total of ∼470,000 residents. All consecutive patients with head CT performed for acute head trauma (n=3010) formed the initial population of this study. Criteria for treatment, and indication for acute head CT, were based on the Scandinavian guidelines for initial management of minimal, mild, and moderate brain injuries.⁴⁰ All consecutive patients with head CT performed for acute brain injury were screened to obtain a sample of working age adults without pre-injury medical or mental health problems who had sustained a “pure” MTBI and who probably could be reached for an outcome visit. The enrolment protocol included three inclusion criteria and nine exclusion criteria as described in our previous publication.⁴¹ Subjects were included if they: 1) met MTBI criteria of the World Health Organization's Collaborating Centre for Neurotrauma Task Force,⁶ 2) were between 18 and 60 years of age, and 3) were residents of the hospital district. Subjects were excluded if they had: 1) premorbid neurological problems, 2) prior psychiatric problems, 3) past TBI, 4) regular psychoactive medication use, 5) neurosurgery, 6) problems with vision or hearing, 7) a first language other than Finnish, 8) a time interval between injury and arrival to the ED>72 h, and/or 9) they declined to participate in the study. For the enrolled sample of 75 patients, detailed prospective data collection was performed that included sociodemographics, injury-related data, and clinical information from the ED. In the current study, we focused on MTBI patients with PTA and MRI images suitable for TA. Alcohol intoxication has a confounding effect on PTA and, therefore, patients who were under the influence of alcohol at the time of injury were excluded from our current study. The final study sample included 50 patients with MTBI. The patients with MTBI were compared with age- and gender-matched orthopedically injured control subjects and healthy controls (n=50). Ethics approval for the study was obtained from the Ethical Committee of Pirkanmaa Hospital District, Finland.

Subjects

Of the final 50 patients with MTBI, 31 (62.0%) were men and 19 (38.0%) were women. The mean age was 38.7 years (SD=11.0, median=37.5, min=19.0, max=60.0). Of the 50 control subjects, 31 (62.0%) were men and 19 (38.0%) were women. The mean age for the control subjects was 37.8 years (SD=11.2, median=37.5, min=18.0, max=57.0). The most common causes of MTBI were sports (n=11, 22.0%) and car accidents (n=11, 22.0%). The other causes of injury were ground-level falls (n=3, 6.0%), motorcycle accidents (n=5, 10.0%), bicycle accidents (n=7, 14.0%), falls from a height (n=8, 16.0%), and other (e.g., object striking the head; n=5, 10.0%). The mean time interval between injury and acute clinical assessment was 50.7 h (median=37.0, SD=51.4, interquartile range [IQR]=20.1–67.5). The mean duration of PTA was 3.2 h (median=2.0, SD=3.5, IQR=0.5–5.0). LOC was witnessed in 13 (26.0%) of the patients with MTBI and mean duration of LOC was 0.8 min (median=0, SD=2.4, IQR=0–0.25). GCS scores were 14 (n=5, 10.0%) or 15 points (n=45, 90.0%). Of the MTBI patients, 11 (22.0%) injuries were complicated and had traumatic intracranial lesions visible on MRI and 5 of these lesions were also identified on head CT. Of the 11 complicated MTBI patients, 2 (18.2%) had a subdural hemorrhage (SDH), 2 (18.2%) had a cerebral contusion, 5 (45.5%) had diffuse axonal injury (DAI), 1 (9.1%) had a subarachnoid hemorrhage, and 1 (9.1%) patient had a SDH and a DAI.

The control subjects (n=50) consisted of 1) adults with an orthopedic injury evaluated in the ED of Tampere University Hospital, and 2) healthy volunteers from the hospital staff and from their families. Controls were age- and gender-matched with the MTBI patients. All controls underwent a head MRI with the same sequences as the MTBI sample.

Clinical assessment of the patients with MTBI

A broad clinical assessment of the patients (n=50) in the final sample was performed by T.M.L. The assessment included a thorough interview of past health, including diagnosed medical conditions, medication use, brain injury history, alcohol consumption according to the Alcohol Use Disorders Identification Test (AUDIT),⁴² and drug and narcotics abuse history. Injury-related data consisted of time of injury, mechanism of injury, helmet use, and alcohol or narcotics intoxication at the time of injury. Presence and duration of possible LOC and disorientation were evaluated using information given by eyewitnesses and ambulance personnel. The presence and duration of retrograde and anterograde amnesia was assessed using the Rivermead PTA protocol.⁴³ PTA was defined as a time between brain injury and the resumption of normal continuous memory. PTA was recorded in hours. The Rivermead PTA protocol has been developed for clinical use, and has been demonstrated to have reasonable reliability for monitoring the duration of PTA in clinical practice.⁴⁴ GCS scores were collected from ambulance forms (if applicable) and the ED. Clinical assessment included a complete neurological examination (cranial and spinal nerves, coordination, balance, pronator drift, and diadochokinesis).

Neuroimaging

In the ED, a non-contrast head CT was performed with a 64 row CT scanner (GE, Lightspeed VCT, WI) for all consecutive patients with brain injury. Head MRI was done with a 3 Tesla Siemens Trio (Siemens AG Medical Solutions, Erlangen, Germany). The MRI protocol included sagittal T1-weighted three-dimensional inversion recovery (3D IR) prepared gradient echo, axial T2 turbo spin echo, conventional axial and high resolution sagittal fluid-attenuated inversion recovery (FLAIR), axial T2*, axial susceptibility weighted imaging (SWI), and diffusion weighted imaging (DWI) series. For the MTBI sample, head CT was performed in the ED within 60 h, and head MRIs were performed within 10 days after injury. Mean time interval between injury and head CT was 11.0 h (SD=12.5, median=6.8, IQR=2.0–14.8) and between injury and head MRI 5.4 days (SD=2.0, median=5.2, IQR=3.9–7.1). All head CT scans and MRIs were analyzed and systematically coded by two neuroradiologists.

TA

Axial FLAIR images (inversion time [TI] 2216 ms, repetition time [TR] 7000 ms, echo time [TE] 87 ms, field of view [FOV] 199×220 ms, matrix 232×256, slice/gap 4.0/1.2 mm, imaged with a12 channel head matrix coil) were chosen for TA. TA was performed with the software package MaZda (MaZda 4.5, Technical University of Lodz, Institute of Electronics)^45
–47 especially designed for TA by Materka and co-workers as part of the European COST B11 and the following COST B21 programs. Using MaZda software, the following spherical regions of interest (ROIs) were manually positioned bilaterally on appropriate image slices: thalamus, amygdala, and hippocampus (Fig. 1). Four different methods of TA were applied: histogram, co-occurrence matrix (COM), run-length matrix (RLM), and autoregressive model-based methods. Calculated texture features are presented in Table 1. RLM parameters were calculated in four directions: horizontal (0 degrees), vertical (90 degrees), 45 degrees, and 135 degrees. COM parameters were calculated in three distances (1, 2, 3 pixels), four times for each distance (in directions θ=0 degrees, 45 degrees, 90 degrees, and 135 degrees). To prevent the positioning of the head in the images from confounding the results, we calculated the mean of each COM parameter over the four directions, resulting in one parameter set for each distance (1, 2, 3 pixels). The gray level normalization of each ROI to minimize the influence of contrast variation and brightness was performed using a method that normalizes image intensities in the range [μ-3σ, μ+3σ].

FIG. 1.

Regions of interest (ROIs) used in the study. ROIs were manually placed bilaterally on T2 axial FLAIR images for each patient and control. (A) Amygdala, (B) hippocampus, and (C) thalamus. Color image is available at www.liebertpub.com/neu

Table 1.

Texture Parameters Used in Analysis

Histogram

Mean, variance, skewness, kurtosis, percentiles 1%, 10%, 50%, 90%, and 99%

Co-occurrence matrix

Angular second moment, contrast, correlation, sum of squares, inverse difference moment, sum average, sum variance, sum entropy, entropy, difference variance, and difference entropy

Run-length matrix

Run-length nonuniformity, gray-level nonuniformity, long-run emphasis, short-run emphasis, and fraction of image in runs

Autoregressive (AR) model

Theta: model parameter vector, four parameters; sigma: standard deviation of the driving noise

Statistical analyses and classification

Mann–Whitney U tests were used to evaluate the raw TA parameters between MTBI patients and matched controls. Comparisons between age groups and time intervals between injury and MRI groups were done by using the Kruskal–Wallis test. IBM SPSS Statistics 20 (IBM Corp. Armonk, NY) was used to perform all statistical analyses. Odds ratios were calculated with 95% confidence intervals, and the statistical significance level was set to 0.01 for all analyses. Spearman correlation coefficients were used to assess the correlations between the mean values of the four texture parameters and the clinical parameters (duration of PTA, age, gender, and time to MRI). To assess if the possibly differing texture features were suitable and powerful enough for classification, the extracted texture features were evaluated by B11,^46,48 a companion software program to MaZda texture software. B11 allows raw and transformed data vectors classification, and evaluation of the usefulness of texture features calculated to classify different texture classes in image regions. B11 assesses the ability of the selected features to distinguish different texture categories of interest with statistical methods such as raw data analysis (RDA), principal component analysis (PCA), and linear discriminant analysis (LDA). In RDA, the assessment of the selected features is based on the original features selected in the input. PCA and LDA are two commonly used techniques for data classification and dimensionality reduction. In PCA, the texture feature data is decomposed, dimension reduced, and represented with principal components. These principal components are also called the most expressive features because they have the ability to optimally represent sets of data.⁴⁸ The LDA method maximizes the ratio of between-class variance to the within-class variance in any particular data set, thereby guaranteeing maximal separability. The main difference between LDA and PCA is that PCA performs more feature classification and LDA performs data classification. In PCA, the shape and location of the original data sets change when transformed to a different space, whereas LDA does not change the location but only tries to provide more class separability and draw a decision region between the given classes. This method gives more information about the distribution of the feature data. A detailed description of the statistical methods provided by B11 can be found in the publication by Hajek and colleagues.⁴⁵

Results

Effects of age, gender, and time of imaging on TA parameters

The first step was to determine if the time between the injury and MRI imaging had any effect on calculated texture features in different ROIs. The MTBI data (n=50) was stratified into four groups based on time when MRI was performed: 1)≤3 days, 2) >3 to ≤5 days, 3) >5 to ≤7 days, and 4) >7 days after injury. There was no significant difference in texture between these time interval divided subgroups (p>0.01, in all analyses). With MTBI patients, no statistically significant correlation was found between time to MRI and TA parameters (p>0.05, in all analyses). The effect of age on texture features was also evaluated. Both patients (n=50) and controls (n=50) were divided in age groups: 1) ≥18 to<30 years, 2) ≥30 to<40 years, 3) ≥40 to<50 years, and 4) ≥50 years. No significant texture differences were observed between these age groups in patients or controls (p>0.01, in all analyses). No statistically significant correlation was found between age and TA parameters (p>0.05, in all analyses) in patients or controls. Regarding gender differences, some texture differences were observed between men and women in MTBI patients and controls. In the MTBI sample, some texture differences were observed between men (n=31) and women (n=19) in the thalami (Table 2). The differing texture features were entropy-based features and angular second moment, which measure the complexity or homogeneity of the given region. The features measuring complexity got higher values in women than in men (sum entropy: p=0.003, Cohen's d=1.05; difference entropy: p=0.005, d=0.99; entropy: p=0.006, d=0.92; angular second moment: p=0.008, d=0.90; difference entropy (pixel distance 2): p=0.001, d=1.21).

Table 2.

Number of Significantly Differing Texture Features between Male and Female Mild Traumatic Brain Injury (MTBI) Patients in Given Regions

	Co-occurrence (COM)-based (n=33)	Autoregressive model-based (n=5)
Thalamus (right)	5	2
Thalamus (left)	3	0
Amygdala (right)	0	0
Amygdala (left)	1	0
Hippocampus (right)	0	0
Hippocampus (left)	0	0

Comparison of TA findings between controls and patients with MTBI

The evaluation of regional differences between MTBI patients with PTA (n=50) and healthy controls (n=50) was based on the number of TA parameters with statistically significant differences. Statistical analyses showed only minor differences between patients with MTBI and healthy controls. These differences were observed in the right side of thalamus (Table 3). The parameters with statistically significant differences (p<0.01) between patients and controls were based on co-occurrence matrix and autoregressive model. No differences were found between patients with MTBI and healthy controls in the left thalamus, hippocampi, or amygdala. The number of significantly differing texture features between patients with MTBI and controls is presented in Table 3. The right thalamus showed significant difference in only 4 out of 65 calculated texture features. Significantly differing texture features between patients and controls in the right thalamus are shown in Table 4. Statistically significant features were used as input in B11 classification analysis (RDA, PCA, and LDA) in order to evaluate if these feature could be used in classification analyses of the right thalamus between patients and controls. The selected features did not, however, prove to be efficient enough for classification of the region. The classification accuracy was only 60–70%. Illustrations of the results are presented in Figure 2.

FIG. 2.

Distribution of raw data analysis (RDA), principal component analysis (PCA), and linear discriminant analysis (LDA) texture features: (A) RDA, (B) PCA, (C) LDA. (Green 2 corresponds to the right side of thalamus in patients and Red 1 corresponds to right side of thalamus in controls). MEF, most effective features, MDF, most discriminative features. Color image is available at www.liebertpub.com/neu

Table 3.

Number of Significantly Differing Texture Features between Mild Traumatic Brain Injury (MTBI) Patients and Controls

	Co-occurrence (COM)-based (n=33)	Autoregressive model-based (n=5)
Thalamus (right)	2	2
Thalamus (left)	0	0
Amygdala (right)	0	0
Amygdala (left)	0	0
Hippocampus (right)	0	0
Hippocampus (left)	0	0

Table 4.

Significantly Differing Texture Features in the Right Thalamus between Patients (n=50) and Controls (n=50)

		Mean		SD
Texture feature groups	Texture features	MTBI	Controls	MTBI	Controls	p value	Cohen's d
COM features	Correlation	0.31	0.36	0.13	0.15	0.006	0.33
	Sum variance	4497.79	4715.64	464.57	494.45	0.001	0.45
ARM features	Teta 1	0.35	0.41	0.12	0.12	0.002	0.54
	Teta 3	0.43	0.48	0.11	0.099	0.01	0.54

MTBI, mild traumatic brain injury; COM, co-occurrence matrix; ARM, autoregressive model.

The patient data were narrowed down with different criteria in order to test if they had an effect on the texture of selected structures. First, it was tested if the duration of PTA had any effect on the texture features. The four significant TA parameters had no correlation with the duration of PTA (p>0.20, in all analyses). In addition, the patients with PTA of>1 h (n=33) were selected, and the four TA parameters were compared with the age- and gender-matched controls (n=33). The results did not differ from results when all patients were analyzed. Whether the traumatic lesions visually detected by a neuroradiologist had an effect on the four texture features was also tested. The MTBI patients with macroscopic traumatic findings were excluded, and the analysis was performed with the remaining sample of uncomplicated MTBIs (n=39). Age-and gender-matched controls were selected for the patients (n=39), and four significant TA parameters were compared between the patients and controls. The texture parameters still differed significantly, only with a slight change in p values. After this, the patients with an uncomplicated MTBI and PTA of>1 h were selected and TA parameters of these subjects were compared with age-and gender-matched controls. Three out of the selected four features (correlation, sum variance, Teta3) still showed statistically significant difference (p<0.01).

Discussion

Visual information in images is often present in the form of texture. The main hypothesis of TA is that by examining the gray-level transitions in images, it is possible to extract a subset of textural features that will characterize the pathology or disease process of interest. The present study used TA to assess the possible microstructural damage in medial temporal lobes of MTBI patients with definite PTA. The aim was to study whether TA can reveal microstructural changes, which might be connected with the duration of PTA in MTBI patients. To the best of our knowledge, there are no similar previous TA studies.

The current study did not reveal significant structural alterations in the studied areas in patients with MTBIs. No difference in texture features were observed between MTBI patients and controls in amygdalae or hippocampi, and only minor differences were present in the right side of the thalamus. The thalamus is composed of several nuclei that project to well-defined cortical areas. With the exception of the olfactory system, every sensory system includes a thalamic nucleus that contralaterally receives specific sensory signals, and sends them to the associated primary cortical area, and sends back information to different thalamic nuclei.⁴⁹ The nuclei in the anterior part of the thalamus interact with the medial temporal structures. They receive input from the hippocampus directly through the fornix and indirectly through the mammillary bodies. Also, anterior and dorsomedial nuclei of the thalamus project back to the hippocampus through the cingulum. Damage to these thalamic nuclei causes both anterograde and retrograde episodic memory deficits. However, the TA parameter changes in the right side of the thalamus did not correlate with duration of PTA, whether the injury was considered to be complicated or uncomplicated, nor did the differences prove to be powerful enough for further classification analysis using PCA or LDA. It is more likely that because the thalamus is very complex and has 50–60 nuclei, the placement of ROI is the cause of these observed minor differences.

When evaluating whether the time interval between the injury and MRI imaging, age, or gender had any effect on calculated texture features in different ROIs, some differences between men and women were found in the thalamus in both MTBI patients and controls. The differing texture features were entropy-based features and angular second moment, which measure the complexity or homogeneity of the given region. The features measuring complexity revealed higher values in women than in men, indicating that the thalami may be structurally more complex and heterogeneous in women. It has been reported that male brains are more asymmetrical than female brains, even in the region of thalamus,⁵⁰ and that these gender-related textural differences might be related to brain asymmetry.

The strength of our study is the carefully selected and well-documented study sample consisting of 50 MTBI patients and 50 controls. All the patients with MTBI were evaluated by the same physician, and, therefore, inter-evaluator bias was eliminated. Compared with many other TA studies,^27,29,30,34 our sample size of 50 patients and age- and gender-matched 50 controls seemed more than sufficient for revealing the hypothesized differences between the groups. The analyses were conducted in a systematic fashion, and through subanalyses, the patient data were narrowed down with different criteria (duration of PTA, time to MRI, and complicated vs. uncomplicated MTBI) in order to test if they had an effect on the texture of selected structures. The limitations of our study were mostly associated with technical factors of TA. One of the obstacles for the widespread clinical application of TA with MRI is the fact that texture features are somewhat sensitive to the choice of MRI equipment, imaging protocols (i.e., sequence parameters), and spatial resolution.^51,52 Fortunately, in the present study, both patients and controls were imaged with the same MRI machine using the same imaging protocol. TA is a data processing procedure, which consists of ROI selection, feature extraction and selection, and classification. ROI selection plays an important role in TA. The ROI selection approach with MaZda requires manually adjusting the location and size of the ROI. In this study, the size and shape of the ROIs were kept constant in order to minimize the effect of the ROI size or shape. This method is still limited in reproducibility, because the location of the ROI is done manually. Because reproducibility is a very important criterion for a clinical imaging tool, further research that includes intra-observer variability measurements would be preferable. The optimal ROI would be large enough to maximize the number of calculated texture features, but small enough to fit the given region to avoid partial volume effect caused by neighboring structures. Even though the placement of ROIs is done with precision, there might still be slight variation in the gray level distribution in the given region. A second point of limitation is the generalizability of the results. The MTBI study sample was highly selected because of numerous study criteria and, therefore, the results cannot be generalized without caution.

Conclusion

The hippocampus plays a fundamental role in memory formation, especially the acquisition of new memories. In MTBI with considerable duration of PTA, such as in the subgroup in the present study with ≥1 h of PTA, hippocampal integrity could be compromised. According to the present study, however, TA did not reveal microstructural changes between patients and controls in studied regions. It is possible that PTA in most MTBI cases arises from reversible physiological alterations, and metabolic and electrolyte imbalance, with limited structural neuron damage.

Footnotes

Acknowledgments

The authors thank research assistants Anne Simi and Marika Suopanki-Ervasti for their contribution to data collection.

Author Disclosure Statement

Grant Iverson has been reimbursed by the government, professional scientific bodies, and commercial organizations for discussing or presenting research relating to mild TBI and sport-related concussion at meetings, scientific conferences, and symposiums. He has a clinical practice in forensic neuropsychology involving individuals who have sustained mild TBIs. He has received honorariums for serving on research panels that provide scientific peer review of programs. The other authors report no competing or conflicts of interest.

References

Russell

W.R.

, and Smith

(1961). Post-traumatic amnesia in closed head injury. Arch. Neurol., 5, 4–17.

Trotter

(1924). Oncertain minor injuries of the brain: Being the Annual Oration, Medical Society of London. Br. Med. J., 1, 816–819.

Symonds

C.P.

(1937). Mental disorder following head injury: (Section of Psychiatry). Proc. R. Soc. Med., 30, 1081–1094.

Ahmed

, Bierley

, Sheikh

J.I.

, and Date

E.S.

(2000). Post-traumatic amnesia after closed head injury: a review of the literature and some suggestions for further research. Brain Inj., 14, 765–780.

Tittle

, and Burgess

G.H.

(2011). Relative contribution of attention and memory toward disorientation or post-traumatic amnesia in an acute brain injury sample. Brain Inj., 25, 933–942.

Holm

, Cassidy

J.D.

, Carroll

L.J.

, Borg

, and Neurotrauma Task Force on Mild Traumatic Brain Injury of the WHO Collaborating Centre. (2005). Summary of the WHO Collaborating Centre for Neurotrauma Task Force on Mild Traumatic Brain Injury. J. Rehabil. Med., 37, 137–141.

Wilson

B.A.

, Evans

J.J.

, Emslie

, Balleny

, Watson

P.C.

, and Baddeley

A.D.

(1999). Measuring recovery from post traumatic amnesia. Brain Inj., 13, 505–520.

Konigs

, de Kieviet

J.F.

, and Oosterlaan

(2012). Post-traumatic amnesia predicts intelligence impairment following traumatic brain injury: a meta-analysis. J. Neurol. Neurosurg. Psychiatry, 83, 1048–1055.

Ponsford

, Willmott

, Rothwell

, Kelly

A.M.

, Nelms

, and Ng

K.T.

(2004). Use of the Westmead PTA scale to monitor recovery of memory after mild head injury. Brain Inj., 18, 603–614.

10.

Ponsford

, Willmott

, Rothwell

, Cameron

, Kelly

A.M.

, Nelms

, Curran

, and Ng

(2000). Factors influencing outcome following mild traumatic brain injury in adults. J. Int. Neuropsychol. Soc., 6, 568–579.

11.

Ponsford

, Cameron

, Fitzgerald

, Grant

, Mikocka–Walus

, and Schonberger

(2012). Predictors of postconcussive symptoms 3 months after mild traumatic brain injury. Neuropsychology, 26, 304–313.

12.

Poyry

, Luoto

T.M.

, Kataja

, Brander

, Tenovuo

, Iverson

G.L.

, and Ohman

(2013). Acute assessment of brain injuries in ground-level falls. J. Head Trauma Rehabil., 28, 89–97.

13.

Tellier

, Marshall

S. C.

, Wilson

K. G.

, Smith

, Perugini

, and Stiell

I. G.

(2009). The heterogeneity of mild traumatic brain injury: Where do we stand?. Brain Inj., 23, 879–887.

14.

Schonberger

, Ponsford

, Reutens

, Beare

, and O'Sullivan

(2009). The Relationship between age, injury severity, and MRI findings after traumatic brain injury. J. Neurotrauma, 26, 2157–2167.

15.

Metting

, Rodiger

L. A.

, de Jong

B. M.

, Stewart

R. E.

, Kremer

B. P.

, and van der Naalt

(2010). Acute cerebral perfusion CT abnormalities associated with posttraumatic amnesia in mild head injury. J. Neurotrauma, 27, 2183–2189.

16.

Hiekkanen

, Kurki

, Brandstack

, Kairisto

, and Tenovuo

(2009). Association of injury severity, MRI-results and ApoE genotype with 1-year outcome in mainly mild TBI: a preliminary study. Brain Inj., 23, 396–402.

17.

Hamm

R. J.

, Temple

M. D.

, Pike

B. R.

, O'Dell

D. M.

, Buck

D. L.

, and Lyeth

B. G.

(1996). Working memory deficits following traumatic brain injury in the rat. J. Neurotrauma, 13, 317–323.

18.

Squire

L. R.

, and Zola–Morgan

(1991). The medial temporal lobe memory system. Science, 253, 1380–1386.

19.

Zola–Morgan

, Squire

L. R.

, and Amaral

D. G.

(1986). Human amnesia and the medial temporal region: enduring memory impairment following a bilateral lesion limited to field CA1 of the hippocampus. J. Neurosci., 6, 2950–2967.

20.

Reed

J. M.

, and Squire

L. R.

(1997). Impaired recognition memory in patients with lesions limited to the hippocampal formation. Behav. Neurosci., 111, 667–675.

21.

Gold

J. J.

, and Squire

L. R.

(2006). The anatomy of amnesia: neurohistological analysis of three new cases. Learn. Mem., 13, 699–710.

22.

Lah

, and Miller

(2008). Effects of temporal lobe lesions on retrograde memory: a critical review. Neuropsychol. Rev., 18, 24–52.

23.

Shenton

M. E.

, Hamoda

H. M.

, Schneiderman

J. S.

, Bouix

, Pasternak

, Rathi

, Vu

M. A.

, Purohit

M. P.

, Helmer

, Koerte

, Lin

A. P.

, Westin

C. F.

, Kikinis

, Kubicki

, Stern

R. A.

, and Zafonte

(2012). A review of magnetic resonance imaging and diffusion tensor imaging findings in mild traumatic brain injury. Brain Imaging Behav., 6, 137–192.

24.

Lerski

R. A.

, Straughan

, Schad

L. R.

, Boyce

, Bluml

, and Zuna

(1993). MR image texture analysis—an approach to tissue characterization. Magn. Reson. Imaging, 11, 873–887.

25.

Castellano

, Bonilha

, Li

L. M.

, and Cendes

(2004). Texture analysis of medical images. Clin. Radiol., 59, 1061–1069.

26.

Bernasconi

, Antel

S. B.

, Collins

D. L.

, Bernasconi

, Olivier

, Dubeau

, Pike

G. B.

, Andermann

, and Arnold

D. L.

(2001). Texture analysis and morphological processing of magnetic resonance imaging assist detection of focal cortical dysplasia in extra-temporal partial epilepsy. Ann. Neurol., 49, 770–775.

27.

Bonilha

, Kobayashi

, Castellano

, Coelho

, Tinois

, Cendes

, and Li

L. M.

(2003). Texture analysis of hippocampal sclerosis. Epilepsia, 44, 1546–1550.

28.

Harrison

L. C.

, Raunio

, Holli

K. K.

, Luukkaala

, Savio

, Elovaara

, Soimakallio

, Eskola

H. J.

, and Dastidar

(2010). MRI texture analysis in multiple sclerosis: toward a clinical analysis protocol. Acad. Radiol., 17, 696–707.

29.

de Oliveira

M. S.

, Balthazar

M. L.

, D'Abreu

, Yasuda

C. L.

, Damasceno

B. P.

, Cendes

, and Castellano

(2011). MR imaging texture analysis of the corpus callosum and thalamus in amnestic mild cognitive impairment and mild Alzheimer disease. AJNR Am. J. Neuroradiol., 32, 60–66.

30.

de Oliveira

M. S.

, D'Abreu

, Franca

M. C.

Jr , Lopes–Cendes

, Cendes

, and Castellano

(2012). MRI-texture analysis of corpus callosum, thalamus, putamen, and caudate in Machado-Joseph disease. J. Neuroimaging, 22, 46–52.

31.

Zhang

, Traboulsee

, Zhao

, Metz

L. M.

, and Li

D. K.

(2011). Texture analysis differentiates persistent and transient T1 black holes at acute onset in multiple sclerosis: a preliminary study. Mult. Scler., 17, 532–540.

32.

Zhang

(2012). MRI texture analysis in multiple sclerosis. Int. J. Biomed. Imaging, 2012, 762804.

33.

Chang

C. W.

, Ho

C. C.

, and Chen

J. H.

(2012). ADHD classification by a texture analysis of anatomical brain MRI data. Front Syst Neurosci, 6, 66.

34.

, Mauss

, Namer

I. J.

, and Chambron

(2001). Existence of contralateral abnormalities revealed by texture analysis in unilateral intractable hippocampal epilepsy. Magn. Reson. Imaging, 19, 1305–1310.

35.

Schad

L. R.

, Bluml

, and Zuna

(1993). MR tissue characterization of intracranial tumors by means of texture analysis. Magn. Reson. Imaging, 11, 889–896.

36.

Mahmoud–Ghoneim

, Toussaint

, Constans

J. M.

, and de Certaines

J. D.

(2003). Three dimensional texture analysis in MRI: a preliminary evaluation in gliomas. Magn. Reson. Imaging, 21, 983–987.

37.

, Mauss

, Zollner

, Namer

I. J.

, and Chambron

(1999). Distinct patterns of active and non-active plaques using texture analysis on brain NMR images in multiple sclerosis patients: preliminary results. Magn. Reson. Imaging, 17, 1261–1267.

38.

Holli

K. K.

, Waljas

, Harrison

, Liimatainen

, Luukkaala

, Ryymin

, Eskola

, Soimakallio

, Ohman

, and Dastidar

(2010). Mild traumatic brain injury: tissue texture analysis correlated to neuropsychological and DTI findings. Acad. Radiol., 17, 1096–1102.

39.

Holli

K. K.

, Harrison

, Dastidar

, Waljas

, Liimatainen

, Luukkaala

, Ohman

, Soimakallio

, and Eskola

(2010). Texture analysis of MR images of patients with mild traumatic brain injury. BMC Med. Imaging, 10, 8-2342-10-8.

40.

Ingebrigtsen

, Romner

, and Kock–Jensen

(2000). Scandinavian guidelines for initial management of minimal, mild, and moderate head injuries. The Scandinavian Neurotrauma Committee. J Trauma, 48, 760–766.

41.

Luoto

T. M.

, Tenovuo

, Kataja

, Brander

, Ohman

, and Iverson

(2013). Who gets recruited in mild traumatic brain injury research?. J. Neurotrauma, 30, 11–16.

42.

Saunders

J. B.

, Aasland

O. G.

, Babor

T. F.

, de la Fuente

J. R.

, and Grant

(1993). Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO Collaborative Project on Early Detection of Persons with Harmful Alcohol Consumption—II. Addiction, 88, 791–804.

43.

Shores

E. A.

, Marosszeky

J. E.

, Sandanam

, and Batchelor

(1986). Preliminary validation of a clinical scale for measuring the duration of post-traumatic amnesia. Med. J. Aust., 144, 569–572.

44.

King

N. S.

, Crawford

, Wenden

F. J.

, Moss

N. E.

, Wade

D. T.

, and Caldwell

F. E.

(1997). Measurement of post-traumatic amnesia: how reliable is it?. J. Neurol. Neurosurg. Psychiatry, 62, 38–42.

45.

Hajek

, Dezortova

, Materka

, and Lerski

(2006). Texture Analysis for Magnetic Resonance Imaging. Prague: Med Publishing.

46.

Materka

(2004). Texture analysis methodologies for magnetic resonance imaging. Dialogues Clin. Neurosci., 6, 243–250.

47.

Szczypinski

P. M.

, Strzelecki

, Materka

, and Klepaczko

(2009). MaZda—a software package for image texture analysis. Comput. Methods Programs Biomed., 94, 66–76.

48.

Luoto

T. M.

, Haikonen

, Haapasalo

, Goebeler

, Huhtala

, Erkinjuntti

, and Karhunen

P. J.

(2009). Large vessel cerebral atherosclerosis is not in direct association with neuropathological lesions of Alzheimer's disease. Eur. Neurol., 62, 93–98.

49.

Herrero

M. T.

, Barcia

, and Navarro

J. M.

(2002). Functional anatomy of thalamus and basal ganglia. Childs Nerv. Syst., 18, 386–404.

50.

Kovalev

V. A.

, Kruggel

, and von Cramon

D. Y.

(2003). Gender and age effects in structural brain asymmetry as measured by MRI texture analysis. Neuroimage, 19, 895–905.

51.

Mayerhoefer

M. E.

, Szomolanyi

, Jirak

, Materka

, and Trattnig

(2009). Effects of MRI acquisition parameter variations and protocol heterogeneity on the results of texture analysis and pattern discrimination: an application-oriented study. Med. Phys., 36, 1236–1243.

52.

Waugh

S. A.

, Lerski

R. A.

, Bidaut

, and Thompson

A. M.

(2011). The influence of field strength and different clinical breast MRI protocols on the outcome of texture analysis using foam phantoms. Med. Phys., 38, 5058–5066.