This systematic review successfully describes the potential of DKI parameters in predicting several acute ischemic stroke outcomes, which include motor, tissue, functional, and neuropsychiatric outcomes.

One of the first prediction studies among the 11 studies included in this review was performed by Spampinato et al. in 2017 and was the only study that focused specifically on motor ability. This highlights the limited availability of motor-specific evidence. The authors assessed the value of DKI in assessing and potentially predicting motor outcomes at 3 months poststroke and reported that DKI-derived metrics are more sensitive than DTI in assessing early cortico-spinal tract (CST) microstructural changes. The outcome assessment tools used were the FMA-UE, mRS, and NIHSS at baseline and at 3-month follow-up, where the acute lesion parameters MK and AK were strongly correlated with the FMA scores. A moderate correlation was also found between lesion volume and the FM-UE score at 3 months poststroke. However, the DKI parameters are more predictive of long-term motor impairment than lesion volume alone and could be potential biomarkers; however, owing to the relatively small sample size, the results require validation. Moreover, only one study has reported motor-specific outcomes, which, in fact, is one of the most common ischemic stroke outcomes. This highlights the need for more longitudinal studies in this context [10].

Functional outcomes are usually evaluated via a mRS, which is an ordered scale coded from 0 (no symptoms at all) to 5 (severe disability) or 6 (death) [21]. One of the first studies to evaluate the correlation between the diffusion dynamics of ischemic lesions and functional outcomes at 6 months poststroke was performed by Liu et al. in 2019. The largest percentage change between the lesion area and the non-lesion area was observed for the MK, and when the lesion volume was controlled, the MK ratio moderately correlated with the functional outcome [12]. Another study assessed not only the lesion but also the DKI parameters in the CST beyond the lesion via voxel-wise and slice-wise analysis instead of typical ROI analysis in the acute stage and functional outcome correlation at 3 months poststroke. They reported that the relative Axial kurtosis (rAK = affected CST - unaffected CST/unaffected CST) independently correlated with the baseline m-NIHSS score (β = 0.297, P = 0.040). However, the long-term correlation between DKI parameters and the mRS score was not significant [13]. The severity of motor impairments is correlated with functional disability and quality of life. However, functional outcome is a broader term for overall outcomes beyond motor outcomes, which could be influenced by a multitude of factors, including rehabilitation, comorbidities, and lifestyle changes [22]. These variables could have perhaps diluted the direct impact of DKI parameters.

Furthermore, Li et al. also assessed CST injury via lesion load analysis which combines DKI-based tract segmentation and orientation mapping, convolutional neural networks, and CST tractograms. It is an innovative technique devoid of subjectivity due to human interaction with increased reproducibility. This study revealed that the MK was significantly correlated with initial motor impairment and functional recovery, with a sensitivity of 0.846 and specificity of 0.682 (p = 0.008). Additionally, the AK had high sensitivity (0.923) but relatively low specificity (0.636) in distinguishing between good and poor outcomes (p < 0.004). These findings highlight the effectiveness of the tractometry-based image processing and analysis techniques [16].

Pei et al. assessed multiple clinical and imaging parameters that could aid as prognostic biomarkers in AIS. The results concerning DKI parameters revealed that the relative Mean Kurtosis (rMK), the ratio between the lesion and contralateral normal value, had the strongest correlation with the mRS score at 90 days postdischarge (r = 0.545, p < 0.001), whereas the relative Mean Diffusivity (rMD) had a negative correlation. ROC analysis revealed that the rMK had the highest predictive accuracy for AIS outcomes, with an AUC of 0.815 [17].

Although motor impairment is a hallmark of post-stroke disability, only one included study directly evaluated motor outcomes, whereas most others relied on global functional scales (e.g., mRS). Our findings, therefore, reflect a broader functional perspective, with limited motor-specific evidence. Future studies incorporating detailed motor assessments are warranted to better delineate the predictive utility of imaging biomarkers for motor recovery.

Currently, Acute ischemic Stroke is imaged using DWI and the corresponding ADC maps; however, these are inadequate for lesion characterization. Although DTI maps further enable imaging the microstructural integrity, the assumption regarding the Gaussian distribution of water molecules is oversimplified. Hence, DKI maps are known to be more sensitive for lesion characterisation as illustrated in Fig. 3.

Representative DWI, DTI, and DKI maps of an 86 y/o female with an acute ischemic stroke in the left parietal lobe. Image courtesy of Department of Radiodiagnosis and Imaging, Kasturba Medical College, Manipal. Patient scanned using United Imaging uMR780 3 T scanner

Further, tissue outcome evaluations reveal whether brain tissue recovers or progresses to infarction. It aids in predicting the extent, number, and location of infarcted tissue on follow-up images. A study by Yin et al. (2018) determined the relationship between DWI and DKI in patients with acute stroke at admission and the tissue outcome by assessing the NIHSS score and stroke volume on T2-weighted MR images one month after stroke onset. Their study revealed that, for large lesions, the acute MK lesion volume was significantly more strongly correlated with the outcome infarction volume than the lesion volume based on DWI parameters (i.e., ADC and MD). In the case of small lesions, the number of kurtosis lesions was persistent on follow-up imaging. Thus, DKI may be a better marker for predicting tissue outcome at 1 month, with respect to the lesion volume and number of lesions. Although the NIHSS score was assessed in this study, no statistical analysis was performed to establish any relationship between the entities [18]. Lampinen et al. also assessed tissue outcomes by performing repeated brain scans at 2-, 9- and 100-day intervals; their findings were in line with previous findings, where the baseline elevated MK values in the lesion implied infarction at the follow-up stage. A high ADC indicates infarction of 50 ± 20% of the lesion volume. Therefore, MK is more sensitive and a better predictor of tissue outcome. However, this study was performed in a very small cohort, and further validation is needed [14].

Early identification and management of neuropsychiatric sequelae, including poststroke depression and cognitive impairment, are essential for improving long-term outcomes in stroke survivors. Poststroke depression (PSD) is a prevalent neuropsychiatric complication associated with an increased risk of impaired functional recovery, diminished quality of life, and elevated mortality that affects nearly one-third of stroke survivors at various stages post-event. [23] Shen et al. (2019) used the Hamilton depression scale with 24 items (HAM-D24) at 3 weeks poststroke to assess the depression status of acute ischemic stroke patients and classified them into three groups: Poststroke Depression (PSD), without poststroke depression (N-PSD), and healthy controls (NORM). Although a direct correlation analysis between HAM-D and DKI parameters was not performed, they did find significant differences in the MK values of the bilateral frontal lobes, temporal lobe, and genu of the corpus callosum of the PSD group compared to those of the N-PSD group and the NORM group, which they suggest could be the underlying mechanism of poststroke depression [11]. Another study in 2021 performed with ROI analysis in the limbic-cortical-striatal-pallidal-thalamic (LCSPT) circuit echoed similar results to those of Shen et al., suggesting that the occurrence of PSD may be due to the decrease in complexity caused by the destruction of white matter microstructure in the frontal lobe, temporal lobe, and genu of the corpus callosum. The values of MK and RK in these regions showed significant negative correlations with HAM-D. These findings strengthen the hypothesis that DKI, especially the MK parameter, could be a potential biomarker for predicting the progression of PSD in early settings but requires validation [15].

Another impact of acute ischemic stroke could manifest as Vascular Cognitive Impairment (VCI), which is a syndrome comprising language, speech, and calculative disability that can worsen over time and even lead to dementia. Hence, its early diagnosis is important [24]. A study by Fan et al. revealed that DKI-derived parameters may be a viable method to evaluate patients with VCI and to predict potential brain structure imaging biomarkers of VCI progress because DKI values in the bilateral frontal lobe, bilateral parietal lobe, genu and splenium of the corpus callosum, anterior and posterior limb of the bilateral internal capsule, bilateral head of the caudate nucleus, bilateral thalamus, and, bilateral medial temporal lobe, are significantly correlated with baseline MMSE scores. Long-term follow-up and correlation analyses were not performed in this study. However, Wu et al. assessed patients longer than in previous studies, i.e., one-year poststroke for imaging and clinical parametric assessment. This study revealed that lower KFA and MK values in white matter tracts were associated with worse cognitive performance, as measured by baseline MMSE and MoCA scores. The participants were then classified as Poststroke Cognitive Impairment (PSCI) and non-cognitive impairment (NCI). Over a year, patients with PSCI presented faster decreases in KFA and MK values, which were correlated with slower cognitive recovery or further decline. This study also revealed that strategic infarct regions, such as the thalamus and frontal lobes, were strongly correlated with reduced KFA and MK values, highlighting their role in cognitive outcomes. Understanding the specific white matter tracts affected helps tailor cognitive rehabilitation programs to improve functions tied to those regions. Thus, allowing clinicians to predict which patients are at a greater risk of PSCI and may require closer follow-up. Even if cognitive functions improve poststroke, persistent white matter damage detected by DKI suggests a need for ongoing monitoring of these patients. This could help in preventing or managing late-onset cognitive decline [20].