Human and AI collaboration for pulmonary nodule segmentation

The article presents Hi-Seg, a human-in-the-loop segmentation framework for pulmonary nodules that integrates the Segment Anything Model (SAM) with human collaboration. In a study involving chest CT scans from 1,179 patients, Hi-Seg achieved a mean Dice score of nearly 85%, surpassing five leading deep learning models by 10-22% and 13 SAM variants by 1-29%. This approach demonstrates the potential to enhance segmentation accuracy while decreasing annotation time, suggesting a transformative impact on clinical workflows and the integration of AI in medical practices.