Using surface electromyography (SEMG) to classify low back pain based on lifting capacity evaluation with principal...

Area of Science:

• Biomedical Engineering
• Clinical Biomechanics
• Rehabilitation Science

Background:

• Low back pain (LBP) is a significant and growing cause of disability worldwide.
• Objective quantitative measures are needed to complement subjective clinical evaluations for LBP.
• Surface electromyography (SEMG) offers a potential objective biomarker for LBP assessment.

Purpose of the Study:

• To differentiate individuals with LBP from healthy subjects using SEMG.
• To establish SEMG-derived quantitative scores for clinical evaluation of LBP.
• To explore the potential of SEMG as a computer-aided diagnosis tool for LBP.

Main Methods:

• Recruited 26 healthy and 26 LBP subjects for static and dynamic lifting tasks with varying weights.
• Extracted multiple features from raw SEMG data, including energy and frequency indexes.
• Utilized false discovery rate (FDR) to remove false positive features and employed principal component analysis neural network (PCANN) for classification.

Main Results:

• Identified specific SEMG features with distinct loadings (30%, 50%) during lifting that differentiate LBP patients from controls.
• Achieved classification accuracies exceeding 80% using the PCANN method across different lifting weights.
• Demonstrated correlations between certain SEMG features and clinical scales measuring exertion, fatigue, and pain.

Conclusions:

• SEMG analysis, particularly with PCANN, provides an objective and accurate method for distinguishing LBP patients.
• The identified SEMG features and achieved accuracy suggest potential for a computer-aided diagnosis tool in LBP evaluation.
• This approach may enhance the quantitative assessment of LBP, correlating objective biomechanical data with subjective clinical experience.