Feature Extraction and Classification of Knee Motions Using a Sensor-Equipped Orthosis

Sammanfattning

This thesis investigates the feasibility of using a sensorized knee orthosis for movementclassification and sensor-derived feature extraction during controlled lowerlimbmovements. The orthosis was equipped with two six-axis IMU sensors and anabsolute capacitive rotary encoder. A movement data-collection study involving 24healthy participants was conducted, in which four lower-limb movements and threeintentionally simulated execution variations were recorded.The collected data were used to train and evaluate three neural network architecturescommonly applied to time-series classification: CNN-LSTM, LSTM-CNN, and aparallel architecture. The study is limited to controlled motion analysis using asensorized wearable system and does not evaluate medical, diagnostic, therapeutic,rehabilitation, or clinical outcomes.The results show that the sensor system provided stable sampling frequencies anddata suitable for classification of four movements. The CNN-LSTM model achieveda mean leave-one-subject-out (LOSO) classification accuracy of 99.42 ± 0.35% forthe predefined movement classes and more than 99% pseudo-live classification accuracyafter post-processing. In addition, movement-related features such as rangeof motion, knee angle deviations, movement smoothness, and repetition count wereextracted from the recorded sensor data. A modified CNN-LSTM architecture wasalso evaluated for classification of predefined execution categories, demonstratingthe feasibility of distinguishing between intentionally defined movement variations.The feature-extraction results should be regarded as exploratory because no externalkinematic ground truth was available. Similarly, the execution-category classificationrepresents predefined movement categories rather than a validated assessment ofmovement quality. Overall, the findings demonstrate the potential of wearable sensorsystems and machine learning for controlled movement classification and sensorderivedmovement analysis.