Hybrid Framework for Cartilage Damage Detection from Vibroacoustic Signals Using Ensemble Empirical Mode...

Area of Science:

• Biomechanics and Biomedical Engineering
• Medical Signal Processing
• Computational Pathology

Background:

• Knee osteoarthritis (OA) is a prevalent degenerative joint disease.
• Early diagnosis of chondromalacia, a key feature of OA, is crucial for effective management.
• Current diagnostic methods can be invasive or lack sensitivity for early-stage cartilage damage.

Purpose of the Study:

• To develop and validate a hybrid analytical framework for non-invasive detection of chondromalacia using vibroacoustic (VAG) signals.
• To compare the performance of different machine learning models (SVM and CNN) in classifying knee OA.
• To assess the potential of VAG signals combined with advanced signal processing and deep learning for early knee pathology diagnosis.

Main Methods:

• Vibroacoustic (VAG) signals were acquired from patients with knee osteoarthritis (OA) and healthy controls (HCs) during knee flexion-extension (open and closed kinetic chain).
• Nonlinear signal decomposition (Ensemble Empirical Mode Decomposition - EEMD) and fluctuation analysis (Detrended Fluctuation Analysis - DFA) were applied to VAG signals.
• Feature extraction, selection (Neighborhood Component Analysis - NCA), and classification using Support Vector Machines (SVM) were performed.
• Convolutional Neural Networks (CNNs) were employed to classify continuous wavelet transform (CWT) scalograms derived from VAG signals.

Main Results:

• The SVM approach achieved high performance in closed kinetic chain (CKC) conditions, with an accuracy of 0.87 and an Area Under the Curve (AUC) of 0.91.
• CNN classification of CWT scalograms demonstrated robust discrimination between OA patients and healthy controls.
• The hybrid framework successfully identified features indicative of cartilage degeneration from VAG signals.

Conclusions:

• The proposed hybrid framework combining multiscale decomposition, nonlinear fluctuation analysis, and deep learning offers an accurate and non-invasive method for detecting cartilage degeneration.
• Vibroacoustic signal analysis holds significant potential for the early diagnosis of knee osteoarthritis and related pathologies.
• The findings support the clinical utility of VAG signals as a diagnostic tool in rheumatology and orthopedics.