Compensatory movement detection by using near-infrared spectroscopy technology based on signal improvement method

Area of Science:

• Biomedical Engineering
• Neurorehabilitation
• Signal Processing

Background:

• Compensatory movements are common in stroke survivors with hemiplegia, hindering recovery.
• Accurate detection of these movements is crucial for effective rehabilitation strategies.

Purpose of the Study:

• To propose and validate a novel method for detecting compensatory movements using near-infrared spectroscopy (NIRS).
• To enhance NIRS signal quality using a differential-based signal improvement (DBSI) method.
• To assess the feasibility of machine learning algorithms for compensatory movement detection.

Main Methods:

• NIRS sensors recorded trunk muscle activation in 10 healthy subjects and 6 stroke survivors during rehabilitation tasks.
• The differential-based signal improvement (DBSI) method was applied to preprocess NIRS signals.
• Support Vector Machine (SVM) algorithm analyzed time-domain features (mean, variance) for compensatory behavior detection.

Main Results:

• NIRS signals demonstrated high accuracy in detecting compensatory movements (97.76% healthy, 97.95% stroke survivors).
• The DBSI method further improved detection accuracy to 98.52% (healthy) and 99.47% (stroke survivors).
• The proposed NIRS-based method showed superior classification performance compared to existing techniques.

Conclusions:

• NIRS technology, combined with DBSI and machine learning, offers a promising, accurate approach for compensatory movement detection in stroke rehabilitation.
• This method has significant potential to improve stroke recovery outcomes and warrants further research.