| 제목 |
Imaging Biomarker-based Model Using for Predicting Upper Extremity Motor Recovery in Stroke Patients |
| 소속 |
Samsung Medical Center, Sungkyunkwan University School of Medicine, Department of Physical and Rehabilitation Medicine, Center for Prevention and Rehabilitation, Heart Vascular Stroke Institute1, Sungkyunkwan University, Departmen of Health Sciences and Technology, Department of Medical Device Management & Research, Department of Digital Health, SAIHST2 |
| 저자 |
Jungsoo Lee1*, Heegoo Kim2, Jinuk Kim2, Won Hyuk Chang1, Yun-Hee Kim1,2† |
Objective
Understanding recovery mechanism and predicting recovery pattern after stroke are important to make individually tailored rehabilitation plans. Various prognostic biomarkers for upper extremity (UE) motor recovery after stroke have been reported. However, most of them had relatively low predictive power in severe stroke patients. This study suggests imaging biomarker-based prediction models for UE motor recovery, including severe stroke patients.
Materials and Methods
Sixty-nine first-ever ischemic stroke patients, who had multimodal imaging data at two weeks and clinical assessments at two weeks and three months after onset, were included. Patients were divided by their initial severity; the severe motor impairment (SMI) group or the mild to moderate motor impairment (MMI) group, and by their recovery pattern based on the proportional recovery rule; Fitter or Non-fitter (i.e., their recovery fit or did not fit the proportional recovery rule). The proportional recovery rule means most patients recover approximately 70% of their initial UE impairment within three to six months. Important neuroimaging biomarkers in motor recovery were investigated. Fractional anisotropy (FA) values of the corticospinal tract (CST), corpus callosum (CC), superior cerebellar peduncle (SCP), lesion volume, lesion load of the CST, interhemispheric homotopic connectivity, and whole brain connectivity were extracted from multimodal imaging data.
Results
All patients with MMI followed the proportional recovery rule. In contrast, patients with SMI showed diverse recovery patterns. Multimodal imaging biomarkers were investigated to improve the predictive accuracy for UE motor recovery in SMI patients. Different imaging biomarkers existed depending on the recovery pattern. The SCP FA was only a significant biomarker in the Fitter. In contrast, the splenium of CC FA was only a significant biomarker in the Non-fitter. Also, the left Heschl’s gyrus (HES) and right superior occipital gyrus (SOG) functional connectivity (HES-SOG FC) was the strongest biomarker among the examined functional connections in the Non-fitter. These biomarkers were reevaluated in patients with different motor-evoked potential (MEP) responses. Finally, the prediction models were established using a stratification strategy according to initial severity and MEP response. The proposed prediction model demonstrated high predictive accuracy (R²=0.855, RMSE=4.37).
Conclusions
Most of all, the imaging biomarker-based prediction model proposed in this study could improve the predictive accuracy of UE motor recovery in patients with SMI. This model may also provide suggests the optimal use of imaging biomarkers for predicting UE motor recovery after stroke. |
|
|
