Using pathological diagnosis as the ground truth, we collected cervical CT scans of patients who had undergone neck lymph node dissection surgery. On these preoperative CT images, we identified and annotated benign and malignant lymph nodes. After annotation, we exported the images as DICOM+RTSS files to ensure precise data handling. The identified regions of interest (ROI) were then isolated for further analysis. This analysis included feature extraction, where we identified specific characteristics of the lymph nodes, mathematical modeling to understand the patterns and relationships within the data, and model prediction to forecast outcomes based on the extracted features and developed models. This comprehensive approach aimed to improve diagnostic accuracy and treatment planning for patients with cervical lymph node involvement.
