A cartilage injury segmentation algorithm based on subordinate degree analysis during lesion location-directed imitation

To accomplish early detection of cartilage injury, avert the ultimate development of degenerative necrosis, and cause permanent harm, MRI images clearly show chondrolesions and overcome the irreversible damage caused by minimally invasive surgery. A cartilage injury segmentation algorithm based on subordinate degree analysis (SDA-LD) is proposed, which can effectively assist clinicians in conducting early diagnosis and determine follow-up plans. On the one hand, to reduce the attention on the non-essential region in knee cartilage images, the location-directed imitation mechanism is required by mimicking the attention allocation process. On the other hand, the proposed SDA-LD method leverages the subordinate degree analysis matrix to determine the association between rich global cartilage information and local lesion texture features. Meanwhile, it properly achieves the correlation between regions of the shape-irregular region of interest (ROI). The experiment operates in the same environment to fairly measure the performance of the proposed algorithm and successfully generates the appropriate quantitative or qualitative analysis by contrasting it with state-of-the-art segmentation algorithms. Experiments show that the proposed SDA-LD algorithm can achieve dice, Jaccard, and recall values of 0.95, 0.90, and 0.95 on real-world medical cartilage injury datasets, respectively.