Fig. 3

KNN Algorithm-Based Knee Joint Pathology Classification Process and Performance Evaluation Across Different K Values. Note: (A) Flowchart of the KNN algorithm classification process, illustrating the complete sequence from calculating Euclidean distances to determining the final classification results. (B) Classification accuracy curves for the training set and testing set at different K values, with the blue line representing the training set and the pink line representing the testing set. (C) Comparison of classification predictions in the training set (a) and testing set (b), where the x-axis represents the knee joint patient samples and the y-axis represents the predicted categories. Red crosses denote the original categories, and blue circles denote the predicted categories. All data were standardized, and the classification results were based on feature extraction and analysis from multimodal MRI data.