On-site Learning for 3D Medical Landmark Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current landmark detection systems in medical imaging are inefficient due to their inability to handle a wide variety of anatomical landmarks, reliance on pre-defined protocols, and limitations in adapting to new imaging modalities, as well as regulatory constraints that prevent off-site training of landmark detector models.
Innovation Solution
A system for automatic on-site learning of landmark detection models that allows end users to define and update landmark detection models based on user feedback, using a database of 3D medical images and machine learning for model generation and updates, enabling fully automatic 3D landmark detection and workflow individualization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual navigation and double oblique alignment are used to detect anatomical landmarks, then detection accuracy can be achieved, but the radiologists' workflow becomes time consuming
Solution Approach 1:
The patent replaces manual mechanical navigation and alignment operations with an automated computer-based system that performs landmark detection and plane orientation automatically, eliminating the time-consuming manual process while maintaining detection accuracy through algorithmic processing
Solution Approach 2:
The system enables self-service by automatically detecting landmarks and orienting planes without requiring radiologist intervention for each step, allowing the system to perform the detection and alignment tasks autonomously based on input images
2Productivity
If standardized algorithms for automatic landmark detection are used, then workflow efficiency is improved, but the system cannot adapt to different clinical questions, modalities, or individual physician preferences
Solution Approach 1:
The patent implements a dynamic system where the landmark detection model can be retrained and updated based on user feedback and new requirements, allowing the system to adapt to different clinical questions, imaging modalities, and individual physician preferences while maintaining automated efficiency
Solution Approach 2:
The system incorporates feedback mechanisms where user interactions and corrections are used to retrain and improve the landmark detection model, enabling continuous adaptation to new requirements and preferences while preserving workflow automation
3Measurement precision
If off-site training of landmark detector models is performed, then model accuracy can be improved, but regulatory constraints prevent data transfer and model updates
Solution Approach 1:
The patent introduces an on-site training capability that acts as an intermediary solution, allowing models to be trained and updated locally within the healthcare institution using their own data, thereby maintaining both accuracy improvement and compliance with regulatory data management requirements
Solution Approach 2:
The system enables self-service by performing model training and updates locally at the institution rather than requiring external off-site training services, allowing the system to maintain high detection accuracy while preserving data sovereignty and meeting regulatory constraints
Data Source
AI summary
A method and system for on-line learning of landmark detection models for end-user specific diagnostic image reading is disclosed. A selection of a landmark to be detected in a 3D medical image is received. A current landmark detection result for the selected landmark in the 3D medical image is determined by automatically detecting the selected landmark in the 3D medical image using a stored landmark detection model corresponding to the selected landmark or by receiving a manual annotation of the selected landmark in the 3D medical image. The stored landmark detection model corresponding to the selected landmark is then updated based on the current landmark detection result for the selected landmark in the 3D medical image. The landmark selected in the 3D medical image can be a set of landmarks defining a custom view of the 3D medical image.