Mild traumatic brain injury (mTBI) is the most common type of traumatic brain injury (TBI), accounting for 75–90% of TBI cases (Mondello et al., 2014; Galgano et al., 2017). Although most mTBI patients are asymptomatic within several days or weeks after injury, 10–30% patients show long-term symptoms, including somatic complaints (headache, dizziness, fatigue, sleep disturbance), behavioral and emotional disorders (anxiety, feeling depressed or tearful, being irritable, reduced self control, loss of initiative and motivation) and subjective cognitive impairment (slowness, memory deficits, concentration difficulties) (Messé et al., 2011).

The diffusion tensor imaging (DTI), which is sensitive to white matter axonal injuries based on Gaussian diffusion, has been used to evaluate fiber tracts damage following mTBI (Wallace et al., 2018; Zhu et al., 2019). These Gaussian diffusion based parameters, including fractional anisotropy (FA), radial diffusivity (RD) and mean diffusivity (MD), which did not consider the non-Gaussian diffusion behavior of in vivo water diffusion within fiber tracts, were measured to reflect the changes of white matter and myelin microstructure after injury. Higher FA, combined with the higher RD and MD, were identified in the mTBI patients when the injury was onset within days, indicating neuronal swelling or cytotoxic edema in the acute stage of the mTBI. Then, the FA decreased to normal during the semi-acute stage in most of the mTBI patients, accompanied with the improvement of post-injury symptoms. While, the athletes with more than one mTBI, combined with the children and young adult populations, had longer increased FA. Furthermore, the patients with persistent post-injury symptoms also had a prolonged period of increased FA. In addition, these patients with persistent post-injury symptoms had lower FA in the chronic stage when compared to healthy controls, and the white matter integrity was correlated with the post-injury symptoms (Chong and Schwedt, 2018). These findings showed loss of myelin and degenerative changes in the mTBI patients with persistent symptoms and demonstrated that the DTI was sensitive in the detection of group-level abnormalities and in the assessment of chronic fiber tract changes.

However, the DTI findings at the group-level may not be applicable for mTBI patient at the individual-level due to the heterogeneity within mTBI patients, which comes from the cause of mTBI (adult civilian, traffic accident, military and sport-related mTBI) and injury position (frontal, temporal, parietal and other position). Moreover, discrepant DTI findings in white matter diffusion metrics were reported across mTBI studies (Asken et al., 2018). For example, high anisotropic diffusion (AD)/low radial diffusivity (RD) in the genu of the corpus callosum was consistently observed in both the replication and original mTBI cohorts (Ling et al., 2012), while another study revealed that acute mTBI was not associated with DTI abnormalities with tract-based spatial statistics in a large and carefully screened sample (Ilvesmäki et al., 2014). The possible reasons for discrepant DTI findings might be control group variability, the mTBI heterogeneity, the analysis techniques, the methods in which the regional differences were reported, and the persistent functional disturbances (Asken et al., 2018). Meanwhile, the complex microstructure, such as crossing fibers, kiss fibers and etc., cannot be resolved by the traditional DTI-based fiber tracking, as well as the exact originations and destinations of fibers (Fernandez-Miranda et al., 2012). Furthermore, the indices in DTI are susceptible to partial volume effects (Oouchi et al., 2007). Additionally, the influence of free-water cannot be removed from the traditional DTI data (Tang et al., 2019).

Recently, differential tractography (Yeh et al., 2019b) and correlation tractography (Yeh et al., 2016a) have been proposed as new tractography modalities to study the white matter tracts. Specifically, differential tractography utilizes repeat MRI scans of the same subjects at different time points to map the exact segment of fiber pathways with neuronal injury in the longitudinal studies. As a quantitative and objective method, it has metrics of monitoring neuronal injury in a single subject without considering inter-subject and group variability, thus allowing for diagnostic and prognostic evaluation of brain diseases at individual-level. Compared to the conventional tractography that maps all the existing pathways, differential tractography could track the precise segments of pathways showing longitudinal changes. Therefore, it has been used to detect neuronal injury on multiple sclerosis, Huntington’s disease, amyotrophic lateral sclerosis, and epileptic patients (Yeh et al., 2019b). The affected pathways shown by differential tractography matched well with the unique clinical symptoms of the patients, and the false discovery rate of the findings could be estimated using a sham setting to provide a reliability measurement. On the other hand, correlation tractography tracks the precise segment of pathways correlated neuropsychological scores in a group of subjects. The results could inform the structure-function relation and explain how white matter affects cognitive functions by disrupting brain circuits. Conventional tractography analysis, which mapped the connections between different brain areas through fiber tracking, has been questioned and its limitations may render track-specific analysis inconclusive. The correlation tractography is more accurate in reflecting the structure and density of the white matter tracts considering crossing fibers and partial volume effects. Hence, it has been applied to map brain connections and correlate findings in neuropsychological disorders (Yeh et al., 2013a; Delaparte et al., 2017; Wang et al., 2019) and neurodegenerative diseases (Mojtahed Zadeh et al., 2018). The differential tractography and correlation tractography have been used to study neuronal changes in multiple brain diseases, but their performances in detecting neuronal injury are yet to be investigated in mTBI patients.

In this study, we aimed to investigate the white matter injury in single-subjects through the fiber tract comparisons between the acute and the follow-up mTBI patients with the differential tractography. Firstly, we performed group comparisons for the acute mTBI patients and the healthy controls with cross sectional differential tractography. Then, the fiber tracts 1 year after injury onset were compared to the white matter of the acute stages from the same patients through the longitudinal differential tractography to evaluate neuronal alterations in single-subject. After that, correlation tractography was performed to find out the exact fiber tracts that correlated with the cognitive behavior. To our knowledge, this is the first study to evaluate trajectory alterations of brain pathways injured from mTBI with differential tractography and correlation tractography, and it can better our understanding in the circuit pathology behind mTBI and the correlation with cognitive manifestations.

This study investigated the white matter alterations and cognitive impairment in the mTBI patients with correlation tractography and differential tractography. The cognitive tests confirmed that the mTBI patients had cognitive impairment in the acute stage. At the same time, the correlation tractography demonstrated that the splenium of the corpus callosum, combined with the right superior longitudinal fasciculus and right cingulum, showed that QA was positively correlated with DSST in the acute mTBI patients. In addition, the network property analysis exhibited decrease in clustering co-efficiency average, small-worldness and global efficiency, and increase of network characteristic path after these correlated fiber bundles were removed. These findings indicated that there maybe an injury in these three fiber bundles at the acute stage of mTBI which resulted in the following cognitive impairments. Furthermore, axonal loss in the corpus callosum was identified in 7 follow-up mTBI patients at individual-level using differential tractography, implying that the potential injury in the corpus callosum resulting from the mTBI may cause axonal loss in the follow-up stage. The potential injury in the right superior longitudinal fasciculus and right cingulum may recover in the follow-up stage, accompanied with better cognitive performance in all the 7 follow-up mTBI patients. Our study demonstrated the cognitive impairment in the acute stage of mTBI, and implied potential axonal loss in the corpus callosum at the chronic stage of mTBI with correlation tractography and differential tractography. Larger sample size longitudinal study is needed to confirm these whiter matter changes in the chronic stage. In addition, the relationship between the QA decrease of injured fiber tracts and the clinical manifestations of individual mTBI patients should be examined in future large study. The differential tractography could be used as a quantitative and objective method to detect neuronal injury in individual mTBI patient.

Previous studies have investigated the fiber bundle changes associated with the cognitive function in mTBI patients through DTI. Lower FA values in specific fiber tracts, including the superior longitudinal fasciculus, cingulum, splenium and genu of corpus callosum, were identified correlating with worse cognitive function (Zhu et al., 2019). The cingulum was mainly associated with the memory and general function, the corpus callosum and superior longitudinal fasciculus mainly participated in the executive function, the corpus callosum was also related to the attention and processing speed (Zhang et al., 2019; Zhu et al., 2019). Moreover, decrease of FA in the superior longitudinal fasciculus, cingulum, and corpus callosum was also identified in the Alzheimer’s disease, multiple sclerosis, amyotrophic lateral sclerosis, Parkinson’s disease and other neurodegenerative diseases and correlated with worse cognitive function in these diseases (Hulst et al., 2013; Zhang et al., 2016; Lo Buono et al., 2020; Wang et al., 2020; Chenji et al., 2021). In our study, the specific fiber tracts identified to correlate with the cognitive function in the correlation tractography, was in line with the findings in previous studies. The higher QA in the splenium of the corpus callosum, right superior longitudinal fasciculus and right cingulum, predicted better cognitive performance in the acute mTBI patients. The cognitive impairment identified in the mTBI patients implied potential injury to these three fiber bundles, though no specific fiber tract was found through group comparisons. The further decreased FA and QA in the corpus callosum, identified among the 7 follow-up mTBI patients using differential tractography, implied potential injury of the corpus callosum after mTBI. The FA, which reflected the diffusivity of the water in the fiber bundle, was good for detecting the integrity of the white matter. The decreased FA in the corpus callosum indicated potential loss of myelin and degenerative changes in this fiber bundle. The QA, which measures the density of anisotropic diffusion water in the fiber bundle, was good for quantifying the amount of the diffusing water along the white matter. The decreased QA in the corpus callosum implied potential changes of the diffusion pattern in this fiber tract (Yeh et al., 2016b). Both decrease of the FA and the QA in the corpus callosum implied potential axonal loss in this fiber tract. Interestingly, all the 7 patients achieved significantly better cognitive performance 1 year after injury, which maybe attributed to the repairmen of the longitudinal fasciculus and cingulum after mTBI. Therefore, we cannot find axonal loss in these two fibers using differential tractography.

Compared to the traditional Gaussian diffusion based DTI, GQI is a model-free method that calculate the SDF directly from diffusion MR signals (Yeh et al., 2010). DTI will result in a large variation in the complexity of the biological changes due to its limitation in the restricted diffusion contributed by axonal myelination. The GQI can reconstruct fibers in complex neuroanatomical regions (Panesar et al., 2019; Li et al., 2021a,b) and tumors surrounding edema zones with accuracy (Zhang et al., 2013; Celtikci et al., 2018). Furthermore, the QA, an anisotropy index calculated from the peak orientations on SDF through GQI, is less sensitive to the partial volume effects of crossing fibers. Additionally, the QA-aided tractography has better resolution than FA-aided tractography and has less false fiber tracks in both shell scheme and grid scheme (Yeh et al., 2013b). Moreover, the QA eliminates the free water effect by removing the isotropic diffusion component and includes difference in the density of diffusion spins so that it is more sensitive. In our presented method, the GQI-based differential tractography was more sensitive in the identification of axonal loss. In the FA differential tractography analysis, corpus callosum axonal loss was only confirmed in Patient 2, Patient 3, and Patient 4, while the axonal loss of corpus callosum in Patient 5 was not significant. At the same time, the axonal loss of corpus callosum in the splenium was not identified in Patient 4 and Patient 7. These findings indicated that the GQI-based differential tractography would be a complementary tool for the DTI in the evaluation of the fiber tract microstructure alterations. Moreover, Patient 3, who developed depression, had a much greater amount of QA and FA decrease in the corpus callosum than other asymptomatic patients, prompting that the symptomatic patients may have much more axonal loss in the chronic stage.

It was important for doctors to assess white matter injury at individual-level since the white matter injury in mTBI patients may be present at the microscopic and molecular level undetectable by structure imaging modality (Shin et al., 2017). However, the previous studies focused on the injury at group level and neglected the inter-subject variability, and cannot be used to evaluate white matter injury at individual-level. Recently, more and more studies focused on the development of a predictive modal to find out potential diffusion biomarkers and to detect mTBI patients based on machine learning (Vergara et al., 2017). As a recent prediction modal that can identify mTBI patients by developing information processing speed deficits with 96.7% accuracy through the combination of DTI indices and inflammation cytokines levels, the frontal-subcortical neuronal circuits would be a potential diffusion predictor for processing speed performance in mTBI patients (Bai et al., 2020). Compared to these prediction modals, the differential tractography would be an indispensable complementary tool to evaluate axonal loss for individual mTBI patients. The differential tractography could not only map the exact injured fiber tracts after mTBI, but also quantify the loss of injured white matter in individuals. As a quantitative and objective method to monitor neuronal alterations in single-subject, it is quite important for the diagnosis and prognosis of mTBI, since many functional and metabolic abnormalities in mTBI may be present in the absence of structural damage. This would provide an objective and quantitative index to evaluate the injury and/or repairmen of white matter for individual mTBI patient.

There are several limitations in our study: (1) Although the alterations of fiber tracts in the chronic stage of mTBI patients were found in our study, longitudinal analysis of larger samples is needed confirm these whiter matter alterations. Furthermore, the axonal alterations should be monitored at multiple time points: including 3 months, 6 months, 1 year, and 2 years after injury. It is necessary to evaluate the injury and/or repairmen of fiber tracts in a dynamic process. (2) Due to the limited number of symptomatic patients in this study, the correlation between the QA decrease of injured fiber tracts and the clinical manifestation should be evaluated in larger sample studies, and the objective differential tractography metrics should be found to reflect the axonal loss of white matter and the clinical symptoms at different stages of individual mTBI patients. (3) The causes of mTBI in our study were heterogeneous, including motor vehicle collision, assault, fall, etc., and the locations of the injury were also various. A homogenous study of mTBI, such as the same location or the same cause of injury would minimize the confounding effect. (4) Only the white matter changes were assessed. Multiple modals should be used in future investigations, including cortical thickness, cortical surface area and cortical volume analysis for cerebral cortex, resting state functional connectivity study for dynamic changes in functional networks, arterial spin labeling technology for cerebral blood flow. Graph theory analysis should also be used in the analysis of spatial relations between brain regions at the global and nodal level. A combination of these modals would be helpful in understanding the injury mechanism of mTBI.

This pilot study demonstrated the cognitive impairment in the acute stage of mTBI. In addition, axonal loss in the corpus callosum was implied at the chronic stage of mTBI patients based on the correlation tractography and differential tractography. These techniques provided a quantitative and objective method to detect neuronal injury in individual mTBI patient, which could be valuable neuroimaging tools to provide clues to the pathophysiological process of white matter alterations in mTBI patients.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.

The studies involving human participants were reviewed and approved by the Ethics Committee of the Second Xiangya Hospital of Central South University. The patients/participants provided their written informed consent to participate in this study.

M-JL and JL contributed to conception and design of the study. S-HH, C-XH, and JL organized the database. M-JL and F-CY contributed to statistical analysis. M-JL and HZ wrote the first draft of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

HZ was employed by the company Siemens Healthcare (China). The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.