Systems and methods for automated tumor segmentation in radiology imaging using data mined line annotations
Patent Information
- Application Number
- EP2023847475
- Authority / Receiving Office
- EP · EP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2022-07-25
- Filing Date
- 2023-07-24
- Publication Date
- 2026-03-04
AI Technical Summary
Manual segmentation of pathological lesions in diagnostic imaging is time-consuming, error-prone, and requires extensive labor, limiting the efficiency and accuracy of detecting and characterizing tumors in radiology.
An AI-driven system that uses unsupervised segmentation techniques to automatically generate and refine pseudo-masks for tumor detection, reducing the need for manually segmented training images by converting line annotations into bounding boxes and iteratively improving segmentation quality through self-refinement and reconciliation processes.
The system enhances the precision of tumor segmentation, decreases the time required for training data encoding, and reduces the reliance on extensive manual annotation, enabling faster and more accurate detection and characterization of tumors in radiology imaging.
Smart Images

Figure 1.1