Artificial intelligence-driven phase recognition in lung surgery: a single-center pilot study

This dataset contains test results from eight cases, used to evaluate an AI-based surgical phase recognition model. Each sample includes per-frame predictions and per-phase summary metrics.Contents1. Frame-level results (eg. resultsW0001.csv)Each row represents the prediction result for a single video frame, with the following columns:image_path: Relative path to the image file (e.g., W0001/x/frame_009600.png)gt_class: Ground truth surgical phase labelpred_class: Predicted phase label by the modelscore: Prediction confidence score (0–1)result: Classification result (e.g., tp, fp, fn, tn)case_number: Case identifier (e.g., W0001)2. Phase-level summary (result_summary.csv)Each row summarizes performance metrics per surgical phase for each case:image_number: Case ID (e.g., W0001)gt_class: Surgical phase label (1 to 5, and X)