Abstract T P30: Novel Algorithm for the Prediction of Motor Recovery of Upper Limb and Finger Using Diffusion Tensor Imaging after Stroke

Abstract

Introduction: Prediction of the outcome of upper limb and finger disability is critical for the rehabilitation in the acute stage of stroke. Here, we show a novel algorithm for the prediction of the recovery patterns by using diffusion tensor imaging (DTI). Methods: 42 supratentorial stroke patients, including both ischemic and hemorrhagic types, with upper limb and finger paralysis were prospectively recruited from May 2012. DTI was performed with a 1.5-T MRI during 14-16 days after onset. Fractional anisotropy (FA) within the cerebral peduncle were measured and the ratios between FA values in the affected and unaffected side (rFA) were calculated. Fiber-tracking was also constructed by the one ROI method using same ROI as the FA measurement in the affected side. DTI findings were categorized into 3 types according to rFA and fiber-tracking (Fig. 1). Brunnstrom stage (BRS, upper limb + finger) was assessed at 2 weeks, 1 month, and 3 months after onset. Recovery patterns were also categorized into 3 groups according to the phase of recovery and final BRS (early recovery: BRS achieved 9-12pt. within 1 month, late recovery: BRS achieved 9-12pt. during 1-3 months, poor recovery: BRS had never achieved 9pt. within 3 months). We statically analyzed the correlation between DTI types and recovery patterns. Results: Positive correlation was found between rFA and final BRS.(p<0.01, Spearman r=0.52) Furthermore, extremely strong correlation was also found between recovery patterns and DTI types.(p<0.0001, Spearman r=0.91) Based on these results, we created a simple algorithm, which can predict the patterns of upper limb and finger recovery during 3 months after stroke with high accuracy. (Fig. 1, early recovery; 73%, late recovery; 85% and poor recovery; 95%) Conclusions: Our algorithm is novel in terms of taking advantages of both FA and fiber-tracking. This method allows us to predict not only final outcome, but also the phase of recovery, which are critical for the rehabilitation.