Prediction of balance function for stroke based on EEG and fNIRS features during ankle dorsiflexion

In what multiverse do you live where predictions of failure to 100% recovery do any good at all for survivor recovery? Create protocols to solve the balance problems post stroke instead of this useless shit. 

Prediction of balance function for stroke based on EEG and fNIRS features during ankle dorsiflexion

Jun Liang^1,2†, Yanxin Song^3†, Abdelkader Nasreddine Belkacem^4*, Fengmin Li¹, Shizhong Liu¹, Xiaona Chen¹, Xinrui Wang¹, Yueyun Wang¹ and Chunxiao Wan^1*

• ¹Department of Rehabilitation, Tianjin Medical University General Hospital, Tianjin, China
• ²Laboratory of Neural Engineering and Rehabilitation, Department of Biomedical Engineering, College of Precision Instruments and Optoelectronics Engineering, Tianjin University, Tianjin, China
• ³Tianjin Tianshi College, Tianjin, China
• ⁴Department of Computer and Network Engineering, College of Information Technology, United Arab Emirates University, Al Ain, United Arab Emirates

Balance rehabilitation is exceedingly crucial during stroke rehabilitation and is highly related to the stroke patients’ secondary injuries (caused by falling). Stroke patients focus on walking ability rehabilitation during the early stage. Ankle dorsiflexion can activate the brain areas of stroke patients, similar to walking. The combination of electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) was a new method, providing more beneficial information. We extracted the event-related desynchronization (ERD), oxygenated hemoglobin (HBO), and Phase Synchronization Index (PSI) features during ankle dorsiflexion from EEG and fNIRS. Moreover, we established a linear regression model to predict Berg Balance Scale (BBS) values and used an eightfold cross validation to test the model. The results showed that ERD, HBO, PSI, and age were critical biomarkers in predicting BBS. ERD and HBO during ankle dorsiflexion and age were promising biomarkers for stroke motor recovery.

Introduction

Stroke is a disease affecting the arteries within the brain, resulting in motor impairment in about 80% of survivors (Langhorne et al., 2009). Among many stroke survivors, most patients were left with sequelae of motor dysfunction, and 30% of patients completely lost the ability to work and became highly disabled (Langhorne et al., 2009; Benjamin et al., 2017). Motor dysfunction causes patients to lose part of their living ability, rendering them unable to complete some daily living activities (Basteris et al., 2014). Therefore, motor recovery always focuses on stroke rehabilitation (Hatem et al., 2016). Balance recovery is essential to motor recovery, as the imbalance-leading falling substantially affects regular training and rehabilitation. In clinical practice, the Berg Balance Scale (BBS) is often used to evaluate the balance function of patients with cerebrovascular and brain injury (Sapmaz and Mujdeci, 2021). However, the scale’s accuracy depends on the experience and subjective judgment of the physical therapists. A biomarker that can illustrate the balance recovery process is necessary to organize the rehabilitation strategy better and improve balance recovery. Developed imaging techniques have given valuable information for diagnostic and functional prognosis. Nevertheless, they may have limitations, such as the special requirements for patients and low temporal resolution (Mukherjee et al., 2008; Buchbinder, 2016). Therefore, more and more studies have concentrated on more convenient methods with electroencephalography (EEG) (Wu et al., 2016; Sebastian-Romagosa et al., 2020).

The EEG acquisition device is simple and portable and has a high temporal resolution. It is highly sensitive to detecting EEG activities and allows subjects to perform some complex limb movement tasks while observing them non-invasively and dynamically in real-time. The neurons’ activity in the brain has been broadly used to monitor the stroke survivors’ brain states (Cillessen et al., 1994; Foreman and Claassen, 2012; Xin et al., 2017). The EEG’s beta band power patterns differed according to the location of the lesion (Park et al., 2016), and event-related desynchronization (ERD) magnitude correlated with residual motor function in the paretic arm (Bartur et al., 2019). However, one challenge of using EEG is its low spatial resolution problem, i.e., the ERD may be contaminated and weakened by the neural activities in the nearby areas. One alternative solution is to use functional near-infrared spectroscopy (fNIRS) as a supplement (Li et al., 2020). In a study using fNIRS to assess the correlation between cortical activation and external postural disturbances, the correlation became stronger with an increase in position-related oxygenated hemoglobin signal and an increase in balance function as measured by the BBS balance scale supplementary motor area (SMA) (Fujimoto et al., 2014). The fNIRS alone has been applied to assess the stroke’s progressive brain plasticity (Delorme et al., 2019). It has also been used with EEG to estimate the effect of different training strategies (Wang et al., 2019). Therefore, combining fNIRS and EEG may give new sight to stroke rehabilitation assessment.

The stroke rehabilitation assessment with EEG or fNIRS was usually undertaken during resting tasks (Nicolo et al., 2015; Sebastian-Romagosa et al., 2020). However, motor recovery should be reflected better during motor or motor imagery tasks (Wang et al., 2019; Li et al., 2020) when the corresponding brain area is activated. Walking ability is an urgent need for stroke patients in the early stage. The assessment should be taken during walking to assess the walking ability of stroke patients precisely. Bipedal locomotion is a complex task requiring maintaining specific motion frequencies, balance and load-bearing, visual integration, and multi-joint coordination (Petersen et al., 2012). However, most stroke survivors during the early stage cannot walk, or they may fall off during walking.

Additionally, ankle dorsiflexion is critical for walking as it occurs throughout the swing phase and at the initiation of the stance phase of gait (Dobkin et al., 2004). How the stroke survivors complete the ankle dorsiflexion affects their walking ability. Therefore, ankle dorsiflexion may be a promising task for stroke rehabilitation assessment (Gennaro and De Bruin, 2020).

This paper aims to evaluate the combination of EEG and fNIRS features during ankle dorsiflexion in rehabilitation assessment. We collected data from stroke survivors during ankle dorsiflexion and built a linear regression model with age, ERD, and oxygenated hemoglobin (HBO) as the predictors and BBS as the response. Our results verified the feasibility of EEG and fNIRS combination in predicting stroke balance state.

More at link.