Localization of early infarction on first-line Non-contrast computed tomogram (NCCT) guides promp... more Localization of early infarction on first-line Non-contrast computed tomogram (NCCT) guides prompt treatment to improve stroke outcome. Our previous study has shown a good performance in the identification of ischemic injury on NCCT. In the present study, we developed a deep learning (DL) localization model to help localize the early infarction sign on NCCT. This retrospective study included consecutive 517 ischemic stroke (IS) patients who received NCCT within 12 h after stroke onset. A total of 21,436 infarction patches and 20,391 non-infarction patches were extracted from the slice pool of 1,634 NCCT according to brain symmetricity property. The generated patches were fed into different pretrained convolutional neural network (CNN) models such as Visual Geometry Group 16 (VGG16), GoogleNet, Residual Networks 50 (ResNet50), Inception-ResNet-v2 (IR-v2), Inception-v3 and Inception-v4. The selected VGG16 model could detect the early infarction in both supratentorial and infratentoria...
ABSTRACT Purpose The objectives are to evaluate the effects of a sequential combination of aerobi... more ABSTRACT Purpose The objectives are to evaluate the effects of a sequential combination of aerobic exercise and cognitive training, compared with exercise or cognitive training alone, on cognitive function, physical function, daily function, quality of life, and social participation in stroke survivors with cognitive impairment. Methods This is a single-blind, parallel, randomized controlled trial. Stroke patients with mild cognitive impairment (n = 56) were randomly assigned to aerobic exercise training (n = 18), computerized cognitive training (n = 18), and the sequential combination of aerobic exercise and computerized cognitive training (n = 20) group. All groups underwent training 60 min/day, 3 days/week, for a total of 12 weeks. The primary outcomes included Montreal Cognitive Assessment (MoCA), Wechsler Memory Scale-Third Edition, and the Stroop color-word test. Secondary outcomes were the Timed Up and Go test, 6-Minute Walk Test, Functional Independence Measure, Lawton Instrumental Activities of Daily Living Scale, Community Integration Questionnaire, and Stroke Impact Scale. Results 56 participants completed the trial. Compared with a single type of aerobic exercise or cognitive training, the combined training group showed significant improvement in MoCA (P < .05, η2 = 0.13), and two sub-tests in WMS-III (both P’s < 0.05) following the intervention. However, no between-group differences were observed for physical functions, daily function, quality of life, and social participation measures. Conclusions The findings provide evidence for the potential synergistic intervention in stroke survivors. Future studies investigating the transfer effects and the optimal training parameters with a larger sample is needed.
Background: Non-invasive techniques such as central intermittent theta burst stimulation (iTBS) a... more Background: Non-invasive techniques such as central intermittent theta burst stimulation (iTBS) and repetitive peripheral magnetic stimulation (rPMS) have shown promise to improve motor function for stroke patients. However, the combined efficacy of rPMS and central iTBS has not been extensively studied. This randomized controlled trial aimed to investigate the synergistic effects of rPMS on central iTBS in patients with stroke. Method: In this study, 28 stroke patients were randomly allocated to receive either 1200 pulses of real or sham rPMS on the radial nerve of the affected limb, followed by 1200 pulses of central iTBS on the ipsilesional hemisphere. The patients received the intervention for 10 sessions over two weeks. The primary outcome measures were the Fugl-Meyer Assessment-Upper Extremity (FMA-UE) and the Action Research Arm Test (ARAT). Secondary outcomes for activities and participation included the Functional Independence Measure-Self care (FIM-Self care) and the Strok...
Early ischemic lesion on non-contrast computed tomogram (NCCT) in acute stroke can be subtle and ... more Early ischemic lesion on non-contrast computed tomogram (NCCT) in acute stroke can be subtle and need confirmation with magnetic resonance (MR) image for treatment decision-making. We retrospectively included the NCCT slices of 129 normal subjects and 546 ischemic stroke patients (onset < 12 h) with corresponding MR slices as reference standard from a prospective registry of Chang Gung Research Databank. In model selection, NCCT slices were preprocessed and fed into five different pre-trained convolutional neural network (CNN) models including Visual Geometry Group 16 (VGG16), Residual Networks 50, Inception-ResNet-v2, Inception-v3, and Inception-v4. In model derivation, the customized-VGG16 model could achieve an accuracy of 0.83, sensitivity 0.85, F-score 0.80, specificity 0.82, and AP 0.82 after using a tenfold cross-validation method, outperforming the pre-trained VGG16 model. In model evaluation, the customized-VGG16 model could correctly identify 53 in 58 subjects (91.37%) ...
Introduction: Compensatory strategies for stroke-related motor impairment may cause a pattern of ... more Introduction: Compensatory strategies for stroke-related motor impairment may cause a pattern of disuse that might limit the improvement of upper limb (UL) motor function, and incomplete UL motor recovery might detrimentally impact quality of life. Previous studies showed that bilateral arm physical training (BAT) can improve UL function after stroke and improves trunk stabilization. However, it remains uncertain that whether BAT can reduce trunk compensation during reaching. Furthermore, there is no study investigating therapist-guided bilateral arm training (TBAT) versus robot-assisted bilateral arm training (RBAT). Hypothesis: We tested the hypothesis that TBAT and RBAT, compared with control treatment (CT), would result in the better outcomes on movement control in the affected UL and less trunk compensation in reaching tasks and would improve quality of life. In addition, TBAT and RBAT may have differential effects on motor control and quality of life. Methods: Forty-two commun...
BACKGROUND Muscle co-contraction during the execution of motor tasks or training is common in pos... more BACKGROUND Muscle co-contraction during the execution of motor tasks or training is common in poststroke subjects. EMG-derived muscular activation indexes have been used to evaluate muscle co-contractions during movements. In addition, robot-assisted bilateral arm training provides a repetitive and stable training method to improve arm movements. However, quantitative measures of muscle contractions during this training in poststroke subjects have not been described. METHODS Seventeen subjects experiencing spastic hemiplegia after a stroke were recruited to perform robot-assisted bilateral wrist flexion and extension movements. The co-contraction index and two new indexes, temporal correlation and cross mutual information, which are derived from the EMGs of working muscles without the need for envelope normalization, are used to quantify intermuscular activation during wrist movements. FINDINGS Higher temporal correlation as well as higher co-contraction index was demonstrated in the affected muscles, implying the recruitment of muscle co-contractions to complete the movement task. On the other hand, a higher value of cross mutual information was exhibited in the unaffected muscles which was attributed to their distinct, rhythmic muscle contractions. The plot of temporal correlation versus cross mutual information further defined affected, unaffected synergistic, and unaffected agonist-antagonist muscular regions. Moreover, with the modified Ashworth scale, multiple regression models based on the co-contraction index and cross mutual information had the highest R-squared value of 0.733. INTERPRETATION EMG-derived intermuscular activation parameters demonstrated muscle co-contractions in the affected muscles and different types of intermuscular contractions during robot-assisted bilateral arm training. The modified Ashworth scale estimation based on multiple regression analysis of the activation indexes also demonstrated EMG-derived index a valuable method for assessing muscle spasticity in subjects with poststroke hemiplegia.
Assessment of upper limb function poststroke is critical for clinical management and determining ... more Assessment of upper limb function poststroke is critical for clinical management and determining the efficacy of interventions. We designed a unilateral upper limb task to simulate activities of daily living to examine how chronic stroke survivors manage reaching, grasping and handling skills simultaneously to perform the functional task using kinematic analysis. The aim of the study was to compare the motor strategies for performing a functional task between paretic and nonparetic arms. Sixteen chronic stroke survivors were instructed to control an ergonomic spoon to transfer liquid from a large bowl to a small bowl using paretic or nonparetic arm. Kinematic data were recorded using a Vicon motion capture system. Outcome measures included movement duration, relative timing, path length, joint excursions, and trial-to-trial variability. Results showed that movement duration, spoon path length, and trunk path length increased significantly when participants used paretic arm to perform the task. Participants tended to reduce shoulder and elbow excursions, and increase trunk excursions to perform the task with paretic arm and altered the relative timing of the task. Although participants used different motor strategies to perform the task with their paretic arms, we did not find the significant differences in trial-to trial variability of joint excursions between paretic and nonparetic arms. The results revealed differences in temporal and spatial aspects of motor strategies between paretic and nonparetic arms. Clinicians should explore the underlying causes of pathological movement patterns and facilitate preferred movement patterns of paretic arm.
European Journal of Physical and Rehabilitation Medicine, 2019
BACKGROUND Virtual reality and interactive video games could decrease the demands on the time of ... more BACKGROUND Virtual reality and interactive video games could decrease the demands on the time of the therapists. However, the cost of a virtual reality system and the requirement for technical support limits the availability of these systems. Commercial exergames are not specifically designed for therapeutic use, most patients with hemiplegic stroke are either too weak to play the games or develop undesirable compensatory movements. AIM To develop Kinect2Scratch games and compare the effects of training with therapist-based training on upper extremity (UE) function of patients with chronic stroke. DESIGN A randomized controlled single-blinded trial. SETTING An outpatient rehabilitation clinic of a tertiary hospital. POPULATIONS Thirty-three patients with chronic hemiplegic stroke. METHODS We developed 8 Kinect2Scratch games. The participants were randomly assigned to either a Kinect2Scratch game group or a therapist-based training group. The training comprised 24 sessions of 30 minutes over 12 weeks. The primary outcome measure was the Fugl-Meyer UE scale and the secondary outcome measures were the Wolf Motor Function Test and Motor Activity Log. Patients were assessed at baseline, after intervention, and at the 3-month follow-up. We used the Pittsburgh participation scale (PPS) to assess the participation level of patients at each training session and an accelerometer to assess the activity counts of the affected UE of patients was used at the 12th and 24th training sessions. RESULTS Seventeen patients were assigned to the Kinect2Scratch group and 16 were assigned to the therapist-based training group. There were no differences between the two groups for any of the outcome measures post-intervention and at the 3-month follow-up (all p>.05). The level of participation was higher in the Kinect2Scratch group than in the therapist-based training group (PPS 5.25vs. 5.00, p=0.112). The total activity counts of the affected UE was significantly higher in the Kinect2Scratch group than in the therapist-based training group (p<.001). CONCLUSIONS Kinect2Scratch game training was feasible, with effects similar to those of therapist-based training on UE function of patients with chronic stroke.
The American journal of occupational therapy : official publication of the American Occupational Therapy Association
This study establishes the concurrent validity, predictive validity, and responsiveness of the Re... more This study establishes the concurrent validity, predictive validity, and responsiveness of the Revised Nottingham Sensation Assessment (rNSA) during rehabilitation for people with stroke. The study recruited 147 patients with stroke. The main assessment used was the rNSA, and outcome measures were the Fugl-Meyer Assessment sensory subscale (FMA-S) and motor subscale (FMA-M) and the Nottingham Extended Activities of Daily Living (NEADL) scale. Correlation coefficients were good to excellent between the rNSA and the FMA-S. The rNSA proprioception measure was a predictor for the FMA-S. The rNSA stereognosis and tactile-pinprick measures for the proximal upper limb were predictors for the FMA-M and the NEADL scale, respectively. Responsiveness was moderate to large for three subscales of the rNSA (standardized response mean = .51-.83). This study may support the concurrent validity, predictive validity, and responsiveness of the rNSA for people with stroke.
PM & R : the journal of injury, function, and rehabilitation, Jan 22, 2016
Robot-assisted therapy (RT) is a promising intervention for stroke rehabilitation. RT hybridized ... more Robot-assisted therapy (RT) is a promising intervention for stroke rehabilitation. RT hybridized with therapist-mediated therapy (eg, RT plus task-specific or impairment-oriented training) may possibly yield functionally relevant improvements. A comparative study of the different combination regimens is needed. To investigate the efficacy of RT combined with task-specific training or impairment-oriented training on motor function and quality of life in patients with chronic stroke. A single-blind, randomized comparative efficacy study. Two medical centers in Taiwan. Twenty-one subjects with chronic stroke. Participants were recruited and randomized into 1 of 2 groups: (1) RT combined with task-specific (RTT) training (enrolled, n = 11; completed, n = 11) or (2) RT combined with impairment-oriented (RTI) training (enrolled, n = 10; completed, n = 9). Participants received 20 intervention sessions (90-100 min/d, 5 d/wk for 4 weeks). The Fugl-Meyer Motor Assessment Upper Extremity subs...
Localization of early infarction on first-line Non-contrast computed tomogram (NCCT) guides promp... more Localization of early infarction on first-line Non-contrast computed tomogram (NCCT) guides prompt treatment to improve stroke outcome. Our previous study has shown a good performance in the identification of ischemic injury on NCCT. In the present study, we developed a deep learning (DL) localization model to help localize the early infarction sign on NCCT. This retrospective study included consecutive 517 ischemic stroke (IS) patients who received NCCT within 12 h after stroke onset. A total of 21,436 infarction patches and 20,391 non-infarction patches were extracted from the slice pool of 1,634 NCCT according to brain symmetricity property. The generated patches were fed into different pretrained convolutional neural network (CNN) models such as Visual Geometry Group 16 (VGG16), GoogleNet, Residual Networks 50 (ResNet50), Inception-ResNet-v2 (IR-v2), Inception-v3 and Inception-v4. The selected VGG16 model could detect the early infarction in both supratentorial and infratentoria...
ABSTRACT Purpose The objectives are to evaluate the effects of a sequential combination of aerobi... more ABSTRACT Purpose The objectives are to evaluate the effects of a sequential combination of aerobic exercise and cognitive training, compared with exercise or cognitive training alone, on cognitive function, physical function, daily function, quality of life, and social participation in stroke survivors with cognitive impairment. Methods This is a single-blind, parallel, randomized controlled trial. Stroke patients with mild cognitive impairment (n = 56) were randomly assigned to aerobic exercise training (n = 18), computerized cognitive training (n = 18), and the sequential combination of aerobic exercise and computerized cognitive training (n = 20) group. All groups underwent training 60 min/day, 3 days/week, for a total of 12 weeks. The primary outcomes included Montreal Cognitive Assessment (MoCA), Wechsler Memory Scale-Third Edition, and the Stroop color-word test. Secondary outcomes were the Timed Up and Go test, 6-Minute Walk Test, Functional Independence Measure, Lawton Instrumental Activities of Daily Living Scale, Community Integration Questionnaire, and Stroke Impact Scale. Results 56 participants completed the trial. Compared with a single type of aerobic exercise or cognitive training, the combined training group showed significant improvement in MoCA (P < .05, η2 = 0.13), and two sub-tests in WMS-III (both P’s < 0.05) following the intervention. However, no between-group differences were observed for physical functions, daily function, quality of life, and social participation measures. Conclusions The findings provide evidence for the potential synergistic intervention in stroke survivors. Future studies investigating the transfer effects and the optimal training parameters with a larger sample is needed.
Background: Non-invasive techniques such as central intermittent theta burst stimulation (iTBS) a... more Background: Non-invasive techniques such as central intermittent theta burst stimulation (iTBS) and repetitive peripheral magnetic stimulation (rPMS) have shown promise to improve motor function for stroke patients. However, the combined efficacy of rPMS and central iTBS has not been extensively studied. This randomized controlled trial aimed to investigate the synergistic effects of rPMS on central iTBS in patients with stroke. Method: In this study, 28 stroke patients were randomly allocated to receive either 1200 pulses of real or sham rPMS on the radial nerve of the affected limb, followed by 1200 pulses of central iTBS on the ipsilesional hemisphere. The patients received the intervention for 10 sessions over two weeks. The primary outcome measures were the Fugl-Meyer Assessment-Upper Extremity (FMA-UE) and the Action Research Arm Test (ARAT). Secondary outcomes for activities and participation included the Functional Independence Measure-Self care (FIM-Self care) and the Strok...
Early ischemic lesion on non-contrast computed tomogram (NCCT) in acute stroke can be subtle and ... more Early ischemic lesion on non-contrast computed tomogram (NCCT) in acute stroke can be subtle and need confirmation with magnetic resonance (MR) image for treatment decision-making. We retrospectively included the NCCT slices of 129 normal subjects and 546 ischemic stroke patients (onset < 12 h) with corresponding MR slices as reference standard from a prospective registry of Chang Gung Research Databank. In model selection, NCCT slices were preprocessed and fed into five different pre-trained convolutional neural network (CNN) models including Visual Geometry Group 16 (VGG16), Residual Networks 50, Inception-ResNet-v2, Inception-v3, and Inception-v4. In model derivation, the customized-VGG16 model could achieve an accuracy of 0.83, sensitivity 0.85, F-score 0.80, specificity 0.82, and AP 0.82 after using a tenfold cross-validation method, outperforming the pre-trained VGG16 model. In model evaluation, the customized-VGG16 model could correctly identify 53 in 58 subjects (91.37%) ...
Introduction: Compensatory strategies for stroke-related motor impairment may cause a pattern of ... more Introduction: Compensatory strategies for stroke-related motor impairment may cause a pattern of disuse that might limit the improvement of upper limb (UL) motor function, and incomplete UL motor recovery might detrimentally impact quality of life. Previous studies showed that bilateral arm physical training (BAT) can improve UL function after stroke and improves trunk stabilization. However, it remains uncertain that whether BAT can reduce trunk compensation during reaching. Furthermore, there is no study investigating therapist-guided bilateral arm training (TBAT) versus robot-assisted bilateral arm training (RBAT). Hypothesis: We tested the hypothesis that TBAT and RBAT, compared with control treatment (CT), would result in the better outcomes on movement control in the affected UL and less trunk compensation in reaching tasks and would improve quality of life. In addition, TBAT and RBAT may have differential effects on motor control and quality of life. Methods: Forty-two commun...
BACKGROUND Muscle co-contraction during the execution of motor tasks or training is common in pos... more BACKGROUND Muscle co-contraction during the execution of motor tasks or training is common in poststroke subjects. EMG-derived muscular activation indexes have been used to evaluate muscle co-contractions during movements. In addition, robot-assisted bilateral arm training provides a repetitive and stable training method to improve arm movements. However, quantitative measures of muscle contractions during this training in poststroke subjects have not been described. METHODS Seventeen subjects experiencing spastic hemiplegia after a stroke were recruited to perform robot-assisted bilateral wrist flexion and extension movements. The co-contraction index and two new indexes, temporal correlation and cross mutual information, which are derived from the EMGs of working muscles without the need for envelope normalization, are used to quantify intermuscular activation during wrist movements. FINDINGS Higher temporal correlation as well as higher co-contraction index was demonstrated in the affected muscles, implying the recruitment of muscle co-contractions to complete the movement task. On the other hand, a higher value of cross mutual information was exhibited in the unaffected muscles which was attributed to their distinct, rhythmic muscle contractions. The plot of temporal correlation versus cross mutual information further defined affected, unaffected synergistic, and unaffected agonist-antagonist muscular regions. Moreover, with the modified Ashworth scale, multiple regression models based on the co-contraction index and cross mutual information had the highest R-squared value of 0.733. INTERPRETATION EMG-derived intermuscular activation parameters demonstrated muscle co-contractions in the affected muscles and different types of intermuscular contractions during robot-assisted bilateral arm training. The modified Ashworth scale estimation based on multiple regression analysis of the activation indexes also demonstrated EMG-derived index a valuable method for assessing muscle spasticity in subjects with poststroke hemiplegia.
Assessment of upper limb function poststroke is critical for clinical management and determining ... more Assessment of upper limb function poststroke is critical for clinical management and determining the efficacy of interventions. We designed a unilateral upper limb task to simulate activities of daily living to examine how chronic stroke survivors manage reaching, grasping and handling skills simultaneously to perform the functional task using kinematic analysis. The aim of the study was to compare the motor strategies for performing a functional task between paretic and nonparetic arms. Sixteen chronic stroke survivors were instructed to control an ergonomic spoon to transfer liquid from a large bowl to a small bowl using paretic or nonparetic arm. Kinematic data were recorded using a Vicon motion capture system. Outcome measures included movement duration, relative timing, path length, joint excursions, and trial-to-trial variability. Results showed that movement duration, spoon path length, and trunk path length increased significantly when participants used paretic arm to perform the task. Participants tended to reduce shoulder and elbow excursions, and increase trunk excursions to perform the task with paretic arm and altered the relative timing of the task. Although participants used different motor strategies to perform the task with their paretic arms, we did not find the significant differences in trial-to trial variability of joint excursions between paretic and nonparetic arms. The results revealed differences in temporal and spatial aspects of motor strategies between paretic and nonparetic arms. Clinicians should explore the underlying causes of pathological movement patterns and facilitate preferred movement patterns of paretic arm.
European Journal of Physical and Rehabilitation Medicine, 2019
BACKGROUND Virtual reality and interactive video games could decrease the demands on the time of ... more BACKGROUND Virtual reality and interactive video games could decrease the demands on the time of the therapists. However, the cost of a virtual reality system and the requirement for technical support limits the availability of these systems. Commercial exergames are not specifically designed for therapeutic use, most patients with hemiplegic stroke are either too weak to play the games or develop undesirable compensatory movements. AIM To develop Kinect2Scratch games and compare the effects of training with therapist-based training on upper extremity (UE) function of patients with chronic stroke. DESIGN A randomized controlled single-blinded trial. SETTING An outpatient rehabilitation clinic of a tertiary hospital. POPULATIONS Thirty-three patients with chronic hemiplegic stroke. METHODS We developed 8 Kinect2Scratch games. The participants were randomly assigned to either a Kinect2Scratch game group or a therapist-based training group. The training comprised 24 sessions of 30 minutes over 12 weeks. The primary outcome measure was the Fugl-Meyer UE scale and the secondary outcome measures were the Wolf Motor Function Test and Motor Activity Log. Patients were assessed at baseline, after intervention, and at the 3-month follow-up. We used the Pittsburgh participation scale (PPS) to assess the participation level of patients at each training session and an accelerometer to assess the activity counts of the affected UE of patients was used at the 12th and 24th training sessions. RESULTS Seventeen patients were assigned to the Kinect2Scratch group and 16 were assigned to the therapist-based training group. There were no differences between the two groups for any of the outcome measures post-intervention and at the 3-month follow-up (all p>.05). The level of participation was higher in the Kinect2Scratch group than in the therapist-based training group (PPS 5.25vs. 5.00, p=0.112). The total activity counts of the affected UE was significantly higher in the Kinect2Scratch group than in the therapist-based training group (p<.001). CONCLUSIONS Kinect2Scratch game training was feasible, with effects similar to those of therapist-based training on UE function of patients with chronic stroke.
The American journal of occupational therapy : official publication of the American Occupational Therapy Association
This study establishes the concurrent validity, predictive validity, and responsiveness of the Re... more This study establishes the concurrent validity, predictive validity, and responsiveness of the Revised Nottingham Sensation Assessment (rNSA) during rehabilitation for people with stroke. The study recruited 147 patients with stroke. The main assessment used was the rNSA, and outcome measures were the Fugl-Meyer Assessment sensory subscale (FMA-S) and motor subscale (FMA-M) and the Nottingham Extended Activities of Daily Living (NEADL) scale. Correlation coefficients were good to excellent between the rNSA and the FMA-S. The rNSA proprioception measure was a predictor for the FMA-S. The rNSA stereognosis and tactile-pinprick measures for the proximal upper limb were predictors for the FMA-M and the NEADL scale, respectively. Responsiveness was moderate to large for three subscales of the rNSA (standardized response mean = .51-.83). This study may support the concurrent validity, predictive validity, and responsiveness of the rNSA for people with stroke.
PM & R : the journal of injury, function, and rehabilitation, Jan 22, 2016
Robot-assisted therapy (RT) is a promising intervention for stroke rehabilitation. RT hybridized ... more Robot-assisted therapy (RT) is a promising intervention for stroke rehabilitation. RT hybridized with therapist-mediated therapy (eg, RT plus task-specific or impairment-oriented training) may possibly yield functionally relevant improvements. A comparative study of the different combination regimens is needed. To investigate the efficacy of RT combined with task-specific training or impairment-oriented training on motor function and quality of life in patients with chronic stroke. A single-blind, randomized comparative efficacy study. Two medical centers in Taiwan. Twenty-one subjects with chronic stroke. Participants were recruited and randomized into 1 of 2 groups: (1) RT combined with task-specific (RTT) training (enrolled, n = 11; completed, n = 11) or (2) RT combined with impairment-oriented (RTI) training (enrolled, n = 10; completed, n = 9). Participants received 20 intervention sessions (90-100 min/d, 5 d/wk for 4 weeks). The Fugl-Meyer Motor Assessment Upper Extremity subs...
Uploads
Papers by Ching-yi Wu