Deep Learning Ultrasound Anatomy Identification for Automatic Exam Setup
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current ultrasound systems require significant time for setup due to numerous controls and settings, and existing automation methods are inadequate for handling the variability in patient anatomy and image quality, limiting the efficiency of exams.
Innovation Solution
Implementing a deep learning neural network model trained on ultrasound images to identify anatomy and adjust system settings automatically, using a neural network to recognize anatomy and its view in real-time, thereby automating the setup process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If ultrasound systems provide numerous controls and settings for different exam types, then the system can handle various exam requirements, but the setup time increases significantly
Solution Approach 1:
The system performs automatic self-setup by analyzing the ultrasound image and autonomously determining exam type, probe type, and optimal settings without requiring manual user configuration. The processor automatically configures all system parameters based on image-based anatomy recognition, eliminating the time-consuming manual setup process while maintaining full adaptability across different exam types.
Solution Approach 2:
The system pre-configures all exam settings, protocols, and parameters in advance based on the automatically recognized anatomy and view type. By determining the appropriate exam type and settings before the actual exam begins, the system eliminates setup time during patient exams while maintaining comprehensive coverage of different exam requirements.
2Measurement precision
If manual setup is required for each exam, then settings can be precisely adjusted, but the complexity of operation increases
Solution Approach 1:
The system automatically determines the precise exam settings, probe type, and exam protocol by analyzing the ultrasound image and recognizing anatomy and view type. This self-configuration process maintains settings accuracy equivalent to manual expert configuration while completely eliminating the operational complexity of manual setup for users.
Solution Approach 2:
The system replaces the manual mechanical process of selecting settings and configuring parameters with an automated image recognition and analysis system. The processor uses deep learning algorithms to automatically determine optimal settings based on visual analysis of the ultrasound image, substituting manual operator actions with automated computational processes that maintain precision while reducing complexity.
3Extent of automation
If deep learning algorithms are used to recognize anatomy, then automation of setup is achieved, but the system complexity increases
Solution Approach 1:
The system employs a single multi-functional deep learning model that performs multiple tasks: recognizing anatomy type, determining view type, and inferring optimal exam settings. This universal model consolidates what would otherwise require separate algorithms and processing systems, achieving high-level automation while minimizing the increase in system complexity through functional integration.
Data Source
AI summary
An ultrasound system with a deep learning neural net feature is used to automatically identify image anatomy or pathology and the view of the anatomy seen in the image. The feature also can assess image quality in real time. Based on identified anatomy, the system can automatically annotate images, launch measurement tools and exam protocols, and perform image control adjustments to aid diagnosis and improve exam workflow.