Potential of GPT-4o for Lung Nodule Feature Characterization and Lung Cancer Probability Estimation from Chest CT Images

This dataset includes (1) 200 lung CT scan videos from the LIDC-IDRI open access dataset, composed of videos from the abdomen to the neck at rates of 3, 6, 12, 20, and 24 frames per second, (2) experimental results based on different frame rates and training strategies on the LIDC-IDRI dataset, (3) experimental results of radiologists' lung RADS scores for 150 cases from the D3 dataset of this study to evaluate the performance of GPT-4o in lung cancer screening, and (4) the test-retest experiment results of GPT-4o on the 135 lung CT scan videos from the LIDC-IDRI open access dataset (radiologists confirmed no lung nodules were found on these CT scans).