Determination of meniscus tears from knee MRI images with artificial intelligence methods

The meniscus is a C-shaped fibrocartilage structure located between the tibia and femur in the knee joint and plays a crucial role in joint stability, load distribution and shock absorption. Damage or degeneration of the meniscus can lead to instability of the joint, impaired load distribution and cartilage wear, which develops into osteoarthritis over time. These changes can lead to restricted movement and chronic pain and significantly impair the patient's quality of life. Meniscus tears are usually diagnosed using magnetic resonance imaging (MRI). However, this method requires expert interpretation, which can affect the diagnostic process and lead to variations in the assessment. The aim of this study is to detect meniscus tears using MRI scans and to precisely segment the meniscus region. To achieve this, a deep learning-based model, MOD-YOLOv8, is proposed for meniscus tear detection. In addition, a novel segmentation approach based on ensemble methods was implemented to improve the accuracy and reliability of meniscus region delineation. YOLO series models are used in the study and the performance of the proposed method is evaluated using metrics such as dice similarity coefficient, precision, recall and positive predictive value. The results show that the proposed model achieves superior performance in both detection and segmentation of meniscal tears, suggesting potential integration into clinical applications.