Multi-Stage Blood Vessel Detection Using Deep Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing blood vessel segmentation technologies are subjective, time-consuming, and limited in accuracy due to the complexity of vascular images, particularly in ill-conditioned angiography, and fail to consider large-scale information effectively.
Innovation Solution
A multi-stage blood vessel detecting method using machine learning algorithms, where the output of one detecting model serves as input for the next, enhancing the accuracy of blood vessel identification in vascular images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual interpretation is used for blood vessel detection, then diagnostic flexibility is maintained, but the process is time-consuming and costly
Solution Approach 1:
The patent replaces manual mechanical interpretation with an automated deep learning system. The U-Net architecture processes vascular images automatically, substituting the manual diagnostic process with computational algorithms that achieve both speed and accuracy in blood vessel segmentation.
Solution Approach 2:
The system performs self-service by automatically processing and interpreting vascular images without requiring manual intervention. The deep learning model independently completes the detection task, though the output is reviewed by doctors for final diagnosis.
2Measurement precision
If existing blood vessel segmentation methods are used, then processing speed is maintained, but accuracy is limited due to inability to handle various blood vessel forms and ill-conditioned angiography
Solution Approach 1:
The patent changes the fundamental parameters of the detection approach by using deep learning with U-Net architecture instead of traditional image processing methods. This involves transforming the input data through multiple convolutional layers with different kernel sizes and activation functions to adapt to various blood vessel forms and conditions.
Solution Approach 2:
The U-Net architecture inherently segments the detection task into multiple stages: encoding paths for feature extraction, skipping connections for spatial information preservation, and decoding paths for segmentation mask generation. This multi-stage segmentation enables accurate handling of diverse blood vessel morphologies.
3Loss of information
If traditional image processing methods are used, then computational resources are reduced, but large-scale information is not considered effectively
Solution Approach 1:
The patent adds the dimension of deep feature learning to traditional 2D image processing. By transforming images through multiple convolutional layers, the system extracts hierarchical features that capture both local details and global contextual information, effectively utilizing large-scale information without excessive computational cost.
Data Source
AI summary
A blood vessel detecting apparatus and an image-based blood vessel detecting method are provided. In the method, first to-be-evaluated data is detected through a first detecting model to obtain a first detection result. Second to-be-evaluated data is detected through a second detecting model to obtain a second detection result. The first to-be-evaluated data includes one or more medical images obtained from photographing a blood vessel. The first detection result output by the first detecting model includes one or more pixels in the medical image belonging to the blood vessel. The first detecting model and the second detecting model are constructed based on a machine learning algorithm. The second to-be-evaluated data includes the first detection result. The second detection result output by the second detecting model includes one or more pixels in the medical image belonging to the blood vessel.