Skip to main content
NCBI home page
Journal of Physical Therapy Science logoLink to Journal of Physical Therapy Science
. 2023 Dec 1;35(12):838–844. doi: 10.1589/jpts.35.838

Applicability of fractional anisotropy from standardized automated tractography for outcome prediction of patients after stroke

Tetsuo Koyama 1,2,*, Midori Mochizuki 1, Yuki Uchiyama 2, Kazuhisa Domen 2
PMCID: PMC10698312  PMID: 38075519

Abstract

[Purpose] Diffusion-tensor fractional anisotropy has been used for outcome prediction in stroke patients. We assessed the clinical applicability of the two major fractional anisotropy methodologies—fractional anisotropy derived from segmentation maps in the standard brain (region of interest) and fractional anisotropy derived from standardized automated tractography—in relation to outcomes. [Participants and Methods] The study design was a retrospective survey of medical records collected from October 2021 to September 2022. Diffusion-tensor imaging was conducted in the second week after stroke onset. Outcomes were assessed using the total score of the motor component of the Stroke Impairment Assessment Set (null to full, 0 to 25). Correlations between fractional anisotropy and the outcomes were then assessed. [Results] Fourteen patients with hemorrhagic stroke were sampled. The fractional anisotropy from standardized automated tractography of the corticospinal tract on the lesion side (mean ± standard deviation, 0.403 ± 0.070) was significantly and tightly correlated (r=0.813) with the outcomes (13.4 ± 9.2), whereas the fractional anisotropy from a region of interest set in the cerebral peduncle on the lesion side (0.548 ± 0.064) was not significantly correlated with the outcomes (r=0.507). [Conclusion] The findings suggest that fractional anisotropy derived from standardized automated tractography can be more applicable to outcome prediction than that derived from a region of interest defined in the standard brain.

Keywords: Prediction, Prognosis, Tractography

INTRODUCTION

Stroke is a major target of rehabilitation medicine worldwide, and outcome prediction is essential for the scheduling appropriate rehabilitative treatment1). Several neuroimaging methods have been proposed for outcome prediction, including magnetic resonance imaging (MRI)2). In particular, diffusion-tensor imaging (DTI) is a relatively new MRI sequence that enables the assessment of neural fiber integrity in vivo3).

DTI requires several different directions of magnetic fields (at least 6) to evaluate the movement of water molecules in brain structures (Fig. 1A)4). In each voxel, the movement of water molecules is estimated as a spheroid (Fig. 1B)5). Of various parameters for these estimated spheroids (Fig. 1B), fractional anisotropy (FA) is most frequently used for the assessment of neural fiber integrity because it is commonly considered an index of Wallerian degeneration due to stroke6).

Fig. 1.

Fig. 1.

Open in a new tab

Schematic illustration of diffusion-tensor imaging and its data processing. A. Magnetic fields from different directions (at least 6) are applied to assess the movement of water molecules in voxels within the brain. B. Water molecules are more likely to move along the neural fiber bundles (λ1), whereas their movement is restricted in perpendicular directions (λ2 and λ3). Such movement of water molecules is estimated as a spheroid. FA values are calculated using the formulae shown. C. In each patient, the obtained FA brain map is spatially transformed into the standard brain, which provides segmentation maps of brain structures. D. Fiber tracking is conducted to fit stream lines between seed voxels and target voxels. The detail settings for the stream lines (e.g., numbers of fibers) are determined in standardized automated tractography protocols. FA: fractional anisotropy.

Several methods have been developed for assessing FA values from DTI datasets. The simplest approach is to derive FA values from a voxel-based FA brain map. This can be manually accomplished via visual inspection7). However, to facilitate better reproducibility, the individual FA maps are often transformed into the standard brain (Fig. 1C)8,9,10,11,12). Next, in reference to the segmentation maps implemented in the standard brain13), the FA values in target voxels (regions of interest [ROIs]) are extracted and then averaged (Fig. 2)12). The procedures for spatial transformation require some computational power but, due to recent advances in digital technology, they can be completed within several minutes on regular Macintosh or Windows computers.

Fig. 2.

Fig. 2.

Open in a new tab

Brain images from a representative patient (No. 8, listed in Table 1). Upper panels show CT images obtained at admission to our acute care service. Middle panels show FA images transformed into the standard brain. Areas shown in light blue indicate cerebral peduncles, which were the ROIs in this study. Lower panels show tractography of this patient. Structures shown in red represent corticospinal tracts. FA values are indicated in yellow. CT: computed tomography; DTI: diffusion-tensor imaging; FA: fractional anisotropy; ROI: region of interest.

A more sophisticated method uses fiber tracking14), in which the spheroids in the brain voxels are estimated to fit stream lines (Fig. 1D) and then the bundles of the stream lines are represented as tractography (Fig. 2)15). Recently, a standardized protocol for automated tractography, including the setting of the seed and target voxels and the numbers of stream lines (Fig. 1D), has become available16). This procedure is computationally expensive17). However, due to the computational power provided by modern graphics processing units (GPUs), the procedures for fiber tracking for a single neural bundle (e.g., the corticospinal tract) take just several minutes17). The mean of the FA values of the voxels representing the neural bundle is then calculated (Fig. 1D). This approach may allow the widespread use of tractography to assess damage to neural bundles in daily clinical practice18).

The aim of this study was to compare clinical applicability between FA derived from the standardized automated tractography and that derived from a ROI defined in the standard brain map in terms of motor outcome prediction in stroke patients.

PARTICIPANTS AND METHODS

This study was based on a retrospective survey of medical records in a local community hospital. Patients who were admitted to Nishinomiya Kyoritsu Neurosurgical Hospital due to stroke between October 2021 and September 2022 were entered into our database. Data were collected from a retrospective survey of medical charts. The patients were treated in accordance with the Japanese Guidelines for the Management of Stroke 2021, including the rehabilitative regimen19). In this study, we focused on patients who had intracerebral hemorrhage in the putamen and/or thalamus because they exhibit greater FA decreases in comparison with those with cerebral infarction11). Conventional computed tomography (CT) images were acquired from patients with intracerebral hemorrhage soon after admission20). Informed consent was obtained by the opt-out method via the hospital website, and the study protocol was approved by the Institutional Review Board of Hyogo Medical University (No. 4453).

To minimize confounds arising from differences in pre-stroke health status, we limited our samples to patients who were functionally independent for activities of daily living before stroke11). We also excluded patients who showed subsequent deterioration in neurological manifestation and other medical conditions during acute care. Furthermore, to minimize differences in the rehabilitative regimen, we sampled data only from patients who were transferred to our affiliated long-term rehabilitation facility, Nishinomiya Kyoritsu Rehabilitation Hospital.

DTI scans were typically performed in the second week following admission to our acute care service using a 3.0-Tesla scanner (MAGNETOM Trio; Siemens AG, Erlangen, Germany) equipped with a 32-channel head coil18). To obtain DTI data, we used a single-shot echo-planar imaging sequence in the anterior-to-posterior direction, comprising 30 images with non-collinear diffusion gradients (b=1,500 s/mm2) and one non-DWI scan (b=0 s/mm2). For each patient, we acquired 80 contiguous axial slices with a field of view of 256 mm × 256 mm, an acquisition matrix of 128 × 128, and a slice thickness of 2 mm. The echo time was 96 ms, the repetition time was 10,900 ms, and the flip angle was 90°. To correct for eddy current- and echo-planar imaging-induced distortions, we also obtained two additional non-DWI scans in the anterior-to-posterior direction and two in the posterior-to-anterior direction. Also, to capture the anatomical details of the patients’ brains, T1-weighted images were also acquired using a three-dimensional fast gradient imaging sequence. For each patient, a total of 176 contiguous sagittal slices were acquired with a field of view of 256 mm × 256 mm, an acquisition matrix of 256 × 256, and a slice thickness of 1 mm. The echo time was 1,900 ms, the repetition time was 2.52 ms, and the flip angle was 10°.

Image processing involved the use of MRtrix software21) and the FMRIB Software Library (FSL)22). Firstly, all DTI scans were denoised. The Gibbs ringing artifact was removed using the “dwidenoise” and “mrdegibbs” commands implemented in MRtrix. Subsequently, corrections for eddy current- and echo-planar imaging-induced distortions were conducted using the “topup” and “eddy” commands provided by FSL. Bias field corrections were performed using the “dwibiascorrect” command in MRtrix using the “ants” option. Then, brain masks for data analysis purposes were generated from the bias field-corrected images using the “bet” command in FSL.

Various DTI parameters (Fig. 1B) were estimated using the “ditfit” command implemented in FSL. In this way, FA brain maps were obtained. Individual FA brain maps were then spatially transformed into the standard brain by using the “fnirt” command implemented in FSL. In reference to the segmentation map (JHU White-Matter Labels)13) implemented in FSL, the FA values in the voxels corresponding to the right and left cerebral peduncles, the ROIs for the present study, were extracted and then averaged (Fig. 2). After the preparation stage using the “bedpostx” tool in FSL, fiber tracking was conducted using the “xtract” command in FSL16). Thereby, tractography for pre-determined neural bundles were constructed. The present study focused on the tractography of the right and left corticospinal tracts. Parameter estimates such as FA values and tract volumes were extracted using the “xtract_stats” command in FSL, with a threshold set at 0.01, based on our previous work23, 24).

Motor function impairment in the upper and lower extremities on the side affected by hemiparesis following intracerebral hemorrhage was evaluated using the motor component of the Stroke Impairment Assessment Set (SIAS-motor)25). This assessment comprises five components, which are scored on a scale from null to full (0 to 5): arm, finger, hip, knee, and ankle functions. In this study, the total sum of SIAS-motor scores was calculated for each patient to quantify gross motor function on the paralyzed side20). In addition, the motor component scores of the Functional Independence Measure (FIM)26), ranging from total dependence to full independence (13 to 91), were obtained for each patient. SIAS-motor and FIM-motor scores were assessed every 2 weeks, and data were collected at the time of discharge from our long-term rehabilitation facility. The total length of hospital stay (LOS), which encompassed acute medical care, was also documented20).

For statistical evaluations, we applied Pearson’s correlation test to examine the associations between FA values and the total SIAS-motor score. Most existing DTI studies in this field use FA ratios between the lesion and non-lesion sides7, 27,28,29,30,31). To ensure the reliability of the present dataset, we used the FA ratio as part of our analyses. To facilitate the use of DTI in real-world setting, we applied raw FA values from the lesion side in addition to the FA ratio. All statistical analyses were conducted using the JMP software package (SAS Institute, Cary, NC, USA), and p-values smaller than 0.05 were considered statistically significant. The correlations for FIM-motor and LOS data will be published elsewhere with higher numbers of participants.

RESULTS

Patients’ profiles are summarized in Table 1. In total, 14 patients were included in the study, 8 males and 6 females. They ranged in age from 43 to 86 years, with a median of 64.5 years. Of these 14 patients, 9 had right hemisphere lesions and 5 had left hemisphere lesions. The SIAS-motor score varied from 0 to 25 (median, 15.5), the FIM-motor score ranged from 27 to 91 (median, 78), and the LOS ranged from 51 to 227 days (median, 149 days).

Table 1. Patients’ profiles.

Patient Gender Age Lesion SIAS-motor FIM LOS FA from DTI ROI (CP) FA from DTI tractography (CST)
No. (years)
-motor
Raw score Total Lesion side Non-lesion side Ratio Lesion side Non-lesion side Ratio
1 M 65 R Thal 5-5-5-5-5 25 91 94 0.657 0.644 1.021 0.530 0.612 0.866
2 M 64 L Thal/Put 5-5-4-4-4 22 80 169 0.510 0.518 0.984 0.496 0.510 0.974
3 M 57 R Thal 5-5-5-5-5 25 89 55 0.585 0.607 0.963 0.490 0.606 0.809
4 F 77 L Thal 5-5-5-5-5 25 89 51 0.597 0.591 1.011 0.438 0.557 0.785
5 M 54 R Put 3-2-4-4-4 17 86 203 0.559 0.609 0.917 0.411 0.549 0.749
6 M 69 R Thal 4-3-4-3-4 18 76 76 0.577 0.577 0.978 0.411 0.556 0.739
7 F 80 R Thal 0-1-2-1-0 4 59 176 0.602 0.602 0.922 0.409 0.598 0.684
8 F 86 R Thal/Put 3-2-4-4-4 17 61 227 0.571 0.592 0.964 0.407 0.567 0.717
9 M 65 L Put 2-0-2-2-1 7 85 145 0.575 0.640 0.897 0.400 0.653 0.613
10 M 53 L Put 1-1-3-3-1 9 82 153 0.569 0.631 0.902 0.371 0.600 0.618
11 M 75 R Thal 1-0-1-2-0 4 27 139 0.492 0.609 0.807 0.369 0.601 0.615
12 F 43 R Put 2-1-3-4-4 14 66 162 0.440 0.529 0.830 0.320 0.556 0.575
13 F 61 R Put 0-0-0-0-0 0 39 205 0.519 0.626 0.828 0.307 0.574 0.534
14 F 57 L Put 0-0-0-0-0 0 38 120 0.424 0.566 0.749 0.288 0.610 0.472
Open in a new tab

Patients are sequenced according to lesion-side FA from DTI tractography (highest to lowest). Four patients (5, 10, 12, and 13) underwent surgical removal of hematoma. Raw SIAS-motor scores are sequenced as arm–finger–hip–knee–ankle. CP: cerebral peduncle; CST: corticospinal tract; F: female; FA: fractional anisotropy; DTI: diffusion-tensor imaging; FIM-motor: motor component of the Functional Independence Measure; Lt: left; LOS: length of hospital stay; M: male; Put: putamen; ROI: region of interest; R: right; SIAS-motor: motor component of the Stroke Impairment Assessment Set; Thal: thalamus.

Figure 3 shows the relationships between FA values and long-term motor outcomes assessed by SIAS-motor. The analysis found significant and tight correlations between the FA ratio and total SIAS-motor score for both the ROI data (r=0.846, p<0.001) and the tractography data (r=0.849, p<0.001). For raw FA values on the lesion side, the data from tractography also exhibited a tight correlation with the total SIAS-motor score (r=0.813, p<0.001). In contrast, results from raw FA values in the ROI data did not reach statistical significance (r=0.507, p=0.064).

Fig. 3.

Fig. 3.

Open in a new tab

Results from correlation analyses between total SIAS-motor scores and FA values. Upper panels show the results of FA derived from ROIs and lower panels show those derived from tractography. Left panels show data for raw FA values and right panels show data for the FA ratio between lesion and non-lesion sides. Red lines represent 90% ellipses, solid lines indicate statistically significant findings, and dashed lines represent statistically non-significant findings. SIAS: stroke impairment assessment set; DTI: diffusion-tensor imaging; FA: fractional anisotropy; ROI: region of interest.

DISCUSSION

To facilitate the applicability of DTI to acute stroke rehabilitation settings, we assessed the clinical utility of FA values derived from ROIs (cerebral peduncles) defined in the standard brain map and those derived from the standardized automated tractography (corticospinal tracts) in relation to motor outcomes. For both types, the FA ratio between the lesion and non-lesion sides was closely correlated with motor outcomes (Fig. 3). While raw FA values on the lesion side in automated tractography were strongly correlated with long-term motor functions, those from ROI were not significantly related to these outcomes (Fig. 3).

The present findings indicated that raw FA data from standardized automated tractography on the lesion side were closely correlated with motor function outcomes. This suggests that the raw FA values of the lesioned bundles can be used for outcome prediction18). In this study, we only included patients with first-ever stroke10). However, recurrence is actually common in stroke patients. Existing studies that focused on FA for outcome prediction often used the FA ratio between the lesion and non-lesion sides7,8,9,10,11,12, 27, 28). This procedure partially minimizes inter-individual differences such as a global FA decline due to aging32). As indicated in the present study (Fig. 3), this approach successfully boosted the statistical power. However, the use of the FA ratio reduces the applicability of DTI for patients with recurrent strokes because the FA in the non-lesion side is supposed to be intact12). As shown in Fig. 3, the present finding that raw FA values derived from standardized automated tractography can sufficiently predict long-term outcomes (r=0.813) may offer a clue for the wider applicability of DTI, even in patients with recurrent strokes.

The analytical procedure of standardized automated tractography requires high computational power. It takes approximate 20 h with a regular Macintosh desktop computer without GPU settings18). The “xtract” command in the current version of FSL (6.0.6) allows the use of modern GPU settings on Linux machines33). This command enables estimates of 42 major neural bundles within the brain, including the corticospinal tract, superior longitudinal fasciculus, and uncinate fasciculus16). By taking advantage of a GPU17), all 42 neural bundles can be analyzed in about 40 min. This aspect can facilitate the applicability of the standardized automated tractography in the real-world setting for acute stroke rehabilitation.

In this study, we obtained DTI data from individuals who had experienced hemorrhagic stroke. We took measures to address the potential confounding influence of MRI susceptibility effects caused by hemosiderin released from the hematoma34). Specifically, we selected ROIs in cerebral peduncles that were distinct from the initial lesion sites within the thalamus, putamen, or both9). Similarly, for the corticospinal tracts, the settings for the standardized automated tractography determine the seed voxels in the lower pons and target voxels in the motor cortex16). Both of them were distant from the hematoma. Accordingly, for both of our ROI and tractography methodologies, signal confounds arising from hemosiderin should be minimal9, 18).

There are several limitations to this work. First, we limited the analytical database to hemorrhagic strokes because our previous report indicated that the FA decrease was less evident in ischemic strokes11). However, as indicated by our previous report, the correlations between FA and hemiparesis severity are comparable for both types of strokes11). This observation suggests that the present findings could be applicable to patients with ischemic stroke. Second, to minimize variability in lesion sites, we exclusively selected patients with hemorrhage in the thalamus and/or putamen. However, in the real-world clinical setting, there are considerable numbers of patients with hemorrhage in subcortical structures and sub-tentorial regions, such as the pons10). Further studies that include such patients are needed to elucidate the effectiveness of the methodologies used in this study. Third, the patient cohort was small (n=14). In the preliminary stage, we performed a power analysis to determine the sufficient number of samples35). According to our previous studies of this topic8,9,10), the expected correlation efficient was 0.7 and power analysis indicated that 14 participants were sufficient to reach statistical significance. Even though our sample size was small, it was appropriate for the purpose of this study.

Taken together, the raw FA values of automated tractography are a good predictor of outcomes in patients after stroke.

Funding

This work was supported in part by a Grant-in-Aid for Scientific Research (C) from the Japan Society for the Promotion of Science (JSPS KAKENHI Grant Number JP22K11356) and by a Grant-in-Aid for Transformative Research Areas-Platforms for Advanced Technologies and Research Resources “Advanced Bioimaging Support” (JSPS KAKENHI Grant Number JP22H04926).

Conflict of interest

The authors declare that there are no conflicts of interest.

REFERENCES

  • 1.Kim B, Winstein C: Can neurological biomarkers of brain impairment be used to predict poststroke motor recovery? A systematic review. Neurorehabil Neural Repair, 2017, 31: 3–24. [DOI] [PubMed] [Google Scholar]
  • 2.Boyd LA, Hayward KS, Ward NS, et al. : Biomarkers of stroke recovery: consensus-based core recommendations from the stroke recovery and rehabilitation roundtable. Neurorehabil Neural Repair, 2017, 31: 864–876. [DOI] [PubMed] [Google Scholar]
  • 3.Moura LM, Luccas R, de Paiva JP, et al. : Diffusion tensor imaging biomarkers to predict motor outcomes in stroke: a narrative review. Front Neurol, 2019, 10: 445. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Lebel C, Benner T, Beaulieu C: Six is enough? Comparison of diffusion parameters measured using six or more diffusion-encoding gradient directions with deterministic tractography. Magn Reson Med, 2012, 68: 474–483. [DOI] [PubMed] [Google Scholar]
  • 5.Westin CF, Maier SE, Mamata H, et al. : Processing and visualization for diffusion tensor MRI. Med Image Anal, 2002, 6: 93–108. [DOI] [PubMed] [Google Scholar]
  • 6.Yu C, Zhu C, Zhang Y, et al. : A longitudinal diffusion tensor imaging study on Wallerian degeneration of corticospinal tract after motor pathway stroke. Neuroimage, 2009, 47: 451–458. [DOI] [PubMed] [Google Scholar]
  • 7.Kusano Y, Seguchi T, Horiuchi T, et al. : Prediction of functional outcome in acute cerebral hemorrhage using diffusion tensor imaging at 3T: a prospective study. AJNR Am J Neuroradiol, 2009, 30: 1561–1565. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Koyama T, Tsuji M, Miyake H, et al. : Motor outcome for patients with acute intracerebral hemorrhage predicted using diffusion tensor imaging: an application of ordinal logistic modeling. J Stroke Cerebrovasc Dis, 2012, 21: 704–711. [DOI] [PubMed] [Google Scholar]
  • 9.Koyama T, Tsuji M, Nishimura H, et al. : Diffusion tensor imaging for intracerebral hemorrhage outcome prediction: comparison using data from the corona radiata/internal capsule and the cerebral peduncle. J Stroke Cerebrovasc Dis, 2013, 22: 72–79. [DOI] [PubMed] [Google Scholar]
  • 10.Koyama T, Marumoto K, Uchiyama Y, et al. : Outcome assessment of hemiparesis due to intracerebral hemorrhage using diffusion tensor fractional anisotropy. J Stroke Cerebrovasc Dis, 2015, 24: 881–889. [DOI] [PubMed] [Google Scholar]
  • 11.Koyama T, Koumo M, Uchiyama Y, et al. : Utility of fractional anisotropy in cerebral peduncle for stroke outcome prediction: comparison of hemorrhagic and ischemic strokes. J Stroke Cerebrovasc Dis, 2018, 27: 878–885. [DOI] [PubMed] [Google Scholar]
  • 12.Koyama T, Uchiyama Y, Domen K: Outcome in stroke patients is associated with age and fractional anisotropy in the cerebral peduncles: a multivariate regression study. Prog Rehabil Med, 2020, 5: 20200006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Hua K, Zhang J, Wakana S, et al. : Tract probability maps in stereotaxic spaces: analyses of white matter anatomy and tract-specific quantification. Neuroimage, 2008, 39: 336–347. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Mori S, Crain BJ, Chacko VP, et al. : Three-dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol, 1999, 45: 265–269. [DOI] [PubMed] [Google Scholar]
  • 15.Kunimatsu A, Aoki S, Masutani Y, et al. : Three-dimensional white matter tractography by diffusion tensor imaging in ischaemic stroke involving the corticospinal tract. Neuroradiology, 2003, 45: 532–535. [DOI] [PubMed] [Google Scholar]
  • 16.Warrington S, Bryant KL, Khrapitchev AA, et al. : XTRACT—standardised protocols for automated tractography in the human and macaque brain. Neuroimage, 2020, 217: 116923. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Hernandez-Fernandez M, Reguly I, Jbabdi S, et al. : Using GPUs to accelerate computational diffusion MRI: from microstructure estimation to tractography and connectomes. Neuroimage, 2019, 188: 598–615. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Mochizuki M, Uchiyama Y, Domen K, et al. : Applicability of automated tractography during acute care stroke rehabilitation. J Phys Ther Sci, 2023, 35: 156–162. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Miyamoto S, Ogasawara K, Kuroda S, et al. Committee for Stroke Guideline 2021, the Japan Stroke Society: Japan Stroke Society Guideline 2021 for the treatment of stroke. Int J Stroke, 2022, 17: 1039–1049. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Uchiyama Y, Domen K, Koyama T: Outcome prediction of patients with intracerebral hemorrhage by measurement of lesion volume in the corticospinal tract on computed tomography. Prog Rehabil Med, 2021, 6: 20210050. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Tournier JD, Smith R, Raffelt D, et al. : MRtrix3: a fast, flexible and open software framework for medical image processing and visualisation. Neuroimage, 2019, 202: 116137. [DOI] [PubMed] [Google Scholar]
  • 22.Jenkinson M, Beckmann CF, Behrens TE, et al. : FSL. Neuroimage, 2012, 62: 782–790. [DOI] [PubMed] [Google Scholar]
  • 23.Koyama T, Domen K: Diffusion tensor fractional anisotropy in the superior longitudinal fasciculus correlates with functional independence measure cognition scores in patients with cerebral infarction. J Stroke Cerebrovasc Dis, 2017, 26: 1704–1711. [DOI] [PubMed] [Google Scholar]
  • 24.Koyama T, Uchiyama Y, Domen K: Associations of diffusion-tensor fractional anisotropy and FIM outcome assessments after intracerebral hemorrhage. J Stroke Cerebrovasc Dis, 2018, 27: 2869–2876. [DOI] [PubMed] [Google Scholar]
  • 25.Tsuji T, Liu M, Sonoda S, et al. : The stroke impairment assessment set: its internal consistency and predictive validity. Arch Phys Med Rehabil, 2000, 81: 863–868. [DOI] [PubMed] [Google Scholar]
  • 26.Heinemann AW, Linacre JM, Wright BD, et al. : Relationships between impairment and physical disability as measured by the functional independence measure. Arch Phys Med Rehabil, 1993, 74: 566–573. [DOI] [PubMed] [Google Scholar]
  • 27.Yoshioka H, Horikoshi T, Aoki S, et al. : Diffusion tensor tractography predicts motor functional outcome in patients with spontaneous intracerebral hemorrhage. Neurosurgery, 2008, 62: 97–103, discussion 103. [DOI] [PubMed] [Google Scholar]
  • 28.Stinear CM, Barber PA, Petoe M, et al. : The PREP algorithm predicts potential for upper limb recovery after stroke. Brain, 2012, 135: 2527–2535. [DOI] [PubMed] [Google Scholar]
  • 29.Chang MC, Kwak SG, Park D: Prediction of the motor prognosis with diffusion tensor imaging in hemorrhagic stroke: a meta-analysis. J Integr Neurosci, 2021, 20: 1011–1017. [DOI] [PubMed] [Google Scholar]
  • 30.Schwarz G, Kanber B, Prados F, et al. : Acute corticospinal tract diffusion tensor imaging predicts 6-month functional outcome after intracerebral haemorrhage. J Neurol, 2022, 269: 6058–6066. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Tao J, Li Z, Liu Y, et al. : Performance comparison of different neuroimaging methods for predicting upper limb motor outcomes in patients after stroke. Neural Plast, 2022, 2022: 4203698. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Sullivan EV, Pfefferbaum A: Diffusion tensor imaging and aging. Neurosci Biobehav Rev, 2006, 30: 749–761. [DOI] [PubMed] [Google Scholar]
  • 33.Nemoto K, Dan I, Rorden C, et al. : Lin4Neuro: a customized Linux distribution ready for neuroimaging analysis. BMC Med Imaging, 2011, 11: 3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Haris M, Gupta RK, Husain N, et al. : Measurement of DTI metrics in hemorrhagic brain lesions: possible implication in MRI interpretation. J Magn Reson Imaging, 2006, 24: 1259–1268. [DOI] [PubMed] [Google Scholar]
  • 35.Rousselet GA, Pernet CR: Improving standards in brain-behavior correlation analyses. Front Hum Neurosci, 2012, 6: 119. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Journal of Physical Therapy Science are provided here courtesy of Society of Physical Therapy Science

RESOURCES