Application of Random Forest MR Spinal Image Labeling Based on Multi-Feature Mergence

Magnetic Resonance Imaging (MRI) is one of the important auxiliary methods in the diagnosis of spinal diseases. In this paper, an automatic labeling method of random forest spine based on multi-feature mergence is proposed. Since the coccygeal vertebrae feature is prominent and easy to identify, this paper firstly adopts HOG and GLCM meshing to extract mergence features from the coccygeal vertebrae, and combines the mergence feature with random forest to identify and label the coccygeal vertebrae. On the basis of coccygeal vertebra identification, the HOG feature after 2DPCA is combined with the random forest to identify and label the other vertebra. In addition, the multi-scale sliding window search method is used for image traversal, the redundant information of recognition results is removed by calculating IOU, and the overall sequence image annotation results are optimized by combining the matrix coordinate information between each layer of the magnetic resonance image, which increases the recognition accuracy and improves the robustness of the algorithm. Compared with MLP, SVM and other algorithms, the algorithm adopted in this paper has better recognition accuracy, faster recognition speed, and lower hardware resource consumption.