DESIGN AND DEVELOPMENT OF HUMAN KNEE JOINT MUSCLE(S) CLASSIFICATION SYSTEM USING ARTIFICIAL INTELLIGENCE TECHNIQUE

It is an automatic system (a combination of hardware and software) to examine and evaluate the different leg movements based on sEMG (Surface Electromyogram) sensors blend with state-of-the-art AI algorithms for the classification and recognition of defects in the muscle(s). The proposed system will

Project Title

DESIGN AND DEVELOPMENT OF HUMAN KNEE JOINT MUSCLE(S) CLASSIFICATION SYSTEM USING ARTIFICIAL INTELLIGENCE TECHNIQUE

Project Area of Specialization

Biomedical Engineering

Project Summary

It is an automatic system (a combination of hardware and software) to examine and evaluate the different leg movements based on sEMG (Surface Electromyogram) sensors blend with state-of-the-art AI algorithms for the classification and recognition of defects in the muscle(s). The proposed system will be intelligent enough to classify if there is any abnormality in the knee movement, the system will automatically classify that which muscle(s) are affected. Twelve time-domain signals will be collected for a different range of leg motions. These signals are then sent to MATLAB for further processing and feature extraction. Later the data will be divided into two categories of healthy and unhealthy data. A Support Vector Machine (SVM) which is a supervised machine learning algorithm, will be used for the classification of the signals. After the creation of the model, the app will also be designed using MATLAB App Designer that provides a Graphical User Interface (GUI) to the user. The final testing and performance of the system will be verified by using the k-fold cross-validation test.

Project Objectives

• To design and development of an Intelligent System for the lower limbs muscles to detect the problems in the flexion and extension.
• To construct an algorithm which can automatically classify healthy and unhealthy data based on their muscle movement using the sEMG signal.
• To produce results and report the accuracy using the Support Vector Machine (SVM) for healthy and unhealthy data.

Project Implementation Method

• For data acquisition, we placed four EMG sensors on the Hamstring (semitendinosus, biceps femoris) and Quadriceps (vastus medialis, rectus femoris) muscles of the subject.
• After the placement of the sensors, calibration of the sensors is done to remove the zero error for which we took some garbage values just to ensure that the sensor is working accurately or not.
• Arduino acquired the data from the sensors. To apply AI algorithms, we need to send the data to MATLAB R2019a. We used “MATLAB Support Package for Arduino”, which allows MATLAB to interactively communicate with an Arduino board.
• After that, we will collect data from multiple subjects. The subjects will be asked to perform six exercises which will be demonstrated through a specifically designed GUI.
• In the next phase, we will start work on creating the MATLAB code. We will then extract the features from the signals that will help to classify healthy and unhealthy muscles of the knee joint.

Benefits of the Project

• It can be used in the rehabilitation centres for the automatic classifications of healthy and unhealthy patients based on muscle movements.
• It can also be used in the sports to detect the lower limb injuries of athletes.
• It can also be used by healthcare professionals for the initial diagnosis, that weather the problem is due to muscles or nerves.

Technical Details of Final Deliverable

These are the ten time-domain features which we will extract from the sEMG data:

1. Root Mean Square (RMS)
2. Standard Deviation (SD)
3. Variance (VAR)
4. Integrated EMG (IE)
5. Average Amplitude Change (AAC)
6. Log Detector (LD)
7. Mean Absolute Value (MAV)
8. Maximum Fractal Length (MFL)
9. Simple Square Integral (SSI)
10. Waveform Length (WL).

The specification of the DF-Robot Analog EMG Sensor are:

• Supply Voltage: +3.3V~5.5V
• Operating Voltage: +3.0V
• Output Voltage: 0~3.0V
• Detection Range: +/-1.5mV
• Operating Temperature: 0~50°C
• Weight: 36g
• Module Connector: PH2.0-3P
• Electrode Connector: PJ-342
• Size: 22mm*35mm (0.87inch*1.38inch)
• Wire Length: 50cm (19.69inch)
• Effective spectrum range: 20Hz~500Hz, needs the ADC
• converter which has higher than 8-bit resolution (Arduino 10-bit).

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Medical

Other Industries

Core Technology

Artificial Intelligence(AI)

Other Technologies

Sustainable Development Goals

Good Health and Well-Being for People

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com