An endoscope system based on augmented reality real-time lesion AR labeling method
By using augmented reality technology to preprocess endoscopic video streams and detect lesions, combined with electromagnetic-visual fusion localization and 3D model construction, the problem of accurate localization and annotation of lesions in endoscopic examinations is solved, improving diagnostic efficiency and accuracy.
Patent Information
- Authority / Receiving Office
- CN Β· China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- FIRST HOSPITAL AFFILIATED TO GENERAL HOSPITAL OF PLA
- Filing Date
- 2026-03-24
- Publication Date
- 2026-06-26
AI Technical Summary
There are problems of invalid area interference and motion blur in the real-time video stream of endoscopy. Lesion identification is easily affected by interference objects in the scene. It is difficult to achieve accurate three-dimensional spatial positioning in two-dimensional field of view. Lesion annotation information cannot be effectively integrated with the real-time endoscopy image, which affects diagnostic efficiency.
An augmented reality-based real-time lesion annotation method for endoscopic systems is adopted. By removing invalid regions and repairing motion blur through preprocessing, a two-stage cascaded model is used for lesion detection and segmentation. Electromagnetic-visual fusion positioning is combined to obtain endoscopic pose data, and two-dimensional lesion information is mapped to three-dimensional space to construct a three-dimensional lesion model and adjust it according to tissue deformation. The resulting AR annotation information is then fused with the real-time endoscopic image.
It improves the accuracy of lesion identification and the diagnostic efficiency of endoscopic examination, and enables real-time AR annotation of lesions, providing intuitive visual reference.
Smart Images

Figure CN122290109A_ABST