Medical Image Localization
Overview
Medical image localization studies how to identify where anatomical landmarks, lesions, organs, and devices appear in 2D and 3D scans or clinical videos, supporting diagnosis, measurement, and image-guided procedures. Common approaches include heatmap and coordinate regression, object detection and segmentation, anatomical priors, weakly or self-supervised learning, temporal tracking, registration, and vision-language grounding. Methods are evaluated across X-ray, CT, MRI, ultrasound, pathology, and endoscopy using landmark error, detection and overlap scores, and tests on external sites; some also guide robotic scanning or surgery. A central challenge is that strong average results can mask failures on small or rare targets, across modalities or hospitals, and under strict 3D alignment, while reliable uncertainty and clinical integration remain open problems.
Latest papers
104 papers11 papers in the last four weeks, up 10% on the four weeks before. 0.1% of all new papers.