Ultrasonic Deep Learning Thermal Ablation Region Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional ultrasonic imaging methods struggle to accurately monitor and image thermal ablation regions due to low recognition accuracy, as they rely on bubble formation, which may not occur in all ablation areas and results in poor contrast between ablation and normal tissue, making it difficult to detect the actual thermal damage.
Innovation Solution
A recognition, monitoring, and imaging method based on ultrasonic deep learning that utilizes a convolutional neural network, cyclic neural network, and deep neural network to process ultrasonic radiofrequency data and optical images, correlating spatial positions to enhance the detection of thermal ablation regions without relying on bubble formation, thereby improving recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional ultrasonic imaging method is used for monitoring thermal ablation, then the imaging system is simple and easy to operate, but the recognition accuracy of thermal ablation region is low
Solution Approach 1:
The patent introduces an optical image as an intermediary to bridge the gap between conventional ultrasonic imaging and thermal ablation region recognition. The optical image captures the actual thermal ablation region, which is then used to guide the ultrasonic image recognition process, solving the problem of low recognition accuracy without requiring complete redesign of the imaging system
Solution Approach 2:
The patent creates a copy of the thermal ablation region information by capturing optical images during the ablation process. These optical copies serve as reference data that are integrated with ultrasonic imaging data, enabling accurate identification of the thermal ablation region while maintaining the simplicity of the original ultrasonic imaging system
2Measurement precision
If bubble formation is used as the basis for ultrasonic echo enhancement, then the ultrasonic imaging method is simple to implement, but the imaging accuracy is poor when bubbles are dissipated or not formed
Solution Approach 1:
The patent implements a feedback mechanism where optical images capturing the actual thermal ablation region are continuously acquired and used to correct and update the ultrasonic imaging results. This feedback loop ensures that even when bubbles are dissipated or not formed, the system can still accurately identify the thermal damage area by referring to the optical feedback information
Solution Approach 2:
The patent performs preliminary action by capturing optical images of the thermal ablation region during the ablation process before the bubbles completely dissipate. These preliminary optical records are stored and used as reference data to guide subsequent ultrasonic imaging analysis, ensuring accurate detection even after bubble dissipation
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The method effectively recognizes thermal ablation regions in ultrasonic images with improved accuracy, providing a more precise monitoring and imaging system for thermal ablation processes, independent of temperature changes and bubble formation, thus enhancing clinical effectiveness.
Implementation Method 1
the ultrasonic radiofrequency data point is an ultrasonic scattering echo signal collected by an ultrasonic imaging device
Data Source
AI summary
A recognition, monitoring, and imaging method and system are provided for a thermal ablation region based on ultrasonic deep learning. The recognition method includes obtaining original ultrasonic radiofrequency data, an ultrasonic image, and an optical image during thermal ablation; making ultrasonic radiofrequency data points and pixels in the optical image correspond in a one-to-one manner, and determining a spatial position correspondence between the ultrasonic radiofrequency data points and the optical image; determining a thermal ablation region image according to the spatial position correspondence; constructing a deep learning model according to the thermal ablation region image; superposing a thermal ablation classification image to the ultrasonic image, to determine a to-be-recognized thermal ablation image; and recognizing a thermal ablation region in the ultrasonic image according to the to-be-recognized thermal ablation image. The method and system improve recognition accuracy of the thermal ablation region in the ultrasonic image.