Anterior Cruciate Ligament Tear Detection Based on Deep Convolutional Neural Network.

Area of Science:

• Medical Imaging
• Orthopedics
• Artificial Intelligence

Background:

• Anterior cruciate ligament (ACL) tears are common injuries in athletes, often caused by sudden movements.
• Accurate detection of ACL tears is vital for effective patient treatment and recovery.
• Existing computer vision techniques for ACL tear detection face challenges due to the complex knee ligament structures.

Purpose of the Study:

• To develop an advanced deep learning model for enhanced anterior cruciate ligament (ACL) tear detection.
• To improve the accuracy and distinctiveness of features extracted from knee MRI images for ACL tear diagnosis.

Main Methods:

• A three-layered compact parallel deep convolutional neural network (CPDCNN) was designed and implemented.
• The CPDCNN model was trained and evaluated using the MRNet knee images dataset.
• Performance metrics including accuracy, recall, precision, and F1 score were utilized.

Main Results:

• The proposed CPDCNN achieved an overall accuracy of 96.60% in detecting ACL tears.
• The model demonstrated high performance with a recall rate of 0.9668, precision of 0.9654, and F1 score of 0.9582.
• The CPDCNN model showed superior performance compared to existing state-of-the-art methods for knee tear detection.

Conclusions:

• The developed CPDCNN model offers a highly accurate and effective solution for anterior cruciate ligament (ACL) tear detection in knee MRI images.
• This deep learning approach shows significant potential for clinical application in diagnosing knee injuries.
• The enhanced feature distinctiveness achieved by CPDCNN addresses limitations of previous computer vision methods.