Related Experiment Video
Updated: Feb 5, 2026

Novel Object Recognition Test for the Investigation of Learning and Memory in Mice
Published on: August 30, 2017
Automatic Anatomy Recognition using Neural Network Learning of Object Relationships via Virtual Landmarks
Fengxia Yan1,2, Jayaram K Udupa2, Yubing Tong2
1College of Science, National University of Defense Technology, Changsha 410073, P. R. China.
This study introduces a new virtual landmark (VL) method to improve automatic anatomy recognition (AAR) by better capturing spatial relationships between anatomical objects. The VL neural network approach enhances object localization accuracy in medical imaging.
Area of Science:
- Medical imaging analysis
- Computational anatomy
- Artificial intelligence in healthcare
Background:
- Automatic Anatomy Recognition (AAR) is essential for medical image analysis.
- Current AAR methods struggle with complex parent-offspring spatial relationships.
- Existing AAR relies on geometric centers, limiting accuracy.
Purpose of the Study:
- To develop an improved AAR method using virtual landmarks (VLs).
- To enhance the capture of spatial relationships in anatomical hierarchies.
- To improve object localization accuracy in medical imaging.
Main Methods:
- Developed a one-shot AAR method utilizing VLs.
- Trained neural networks to predict pose and VLs of offspring objects from parent VLs.
- Implemented separate neural networks for VL and pose prediction for parent-offspring pairs.
Main Results:
- The VL-based neural network method demonstrated more accurate object localization.
- Evaluated on 14 objects across two hierarchies using 54 CT datasets.
- Outperformed the traditional AAR method relying on geometric centers.
Conclusions:
- Virtual landmarks significantly improve the accuracy of automatic anatomy recognition.
- The proposed VL neural network method offers a more robust approach to anatomical relationship modeling.
- This advancement has implications for radiation therapy treatment planning and medical image analysis.
Related Concept Videos
Automatic Processing and Automatic Social Behavior
Relationship Formation
Anatomy of the Circulatory System
Virtual Work
In static equilibrium, a body can experience an imaginary or virtual movement, such as displacement or rotation. The virtual work done by a force is equal to the dot product of force and virtual displacement in the direction of the force. When it comes to virtually rotating a...
Ending Relationships
Protein Networks
These interactions can be represented through maps depicting protein-protein interaction networks, represented as nodes and edges. Nodes are circles that are representative of a protein,...

