Blood Flow Spectrum Envelope Extraction from Noisy Ultrasound
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for drawing blood flow spectrum envelopes using computer image recognition are hindered by the complexity of color Doppler ultrasound images, which are sensitive to gain settings, leading to incomplete or fractured displays due to noise and background interference.
Innovation Solution
A method involving image preprocessing with Gaussian smoothing and OTSU threshold segmentation, followed by non-maximum suppression to identify block domains, and contour pixel point tracking to form a blood flow spectrum envelope, using algorithms like Sobel operators to draw the envelope.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If color Doppler ultrasound gain is increased to improve blood flow spectrum visibility, then the brightness of color blocks where blood flow spectra are located is improved, but the background becomes more complex and noise increases, making it difficult to separate color blocks from background
Solution Approach 1:
The patent segments the ultrasound image into different regions by identifying block domains (connected domains) that represent blood flow spectra. This segmentation separates the color blocks containing blood flow information from the complex background, allowing the system to focus processing on relevant regions while filtering out noise and irrelevant background elements.
Solution Approach 2:
The patent extracts the maximum connected domain (block domain) from the ultrasound image that contains the blood flow spectrum information. By taking out this specific region of interest and processing it separately, the system isolates the useful blood flow data from the harmful background complexity and noise, enabling accurate envelope drawing.
2Object-affected harmful factors
If color Doppler ultrasound gain is decreased to simplify background and reduce noise, then the background becomes simpler and noise is reduced, but the brightness of color blocks where blood flow spectra are located becomes lower and parts of color blocks may be fractured, resulting in incomplete display
Solution Approach 1:
The patent performs preliminary actions by applying Gaussian smoothing and threshold processing to the ultrasound image before extracting block domains. These preprocessing steps enhance the visibility and continuity of blood flow spectrum regions, ensuring that even at lower gain settings where the background is simpler, the blood flow spectra remain complete and unfractured, enabling reliable envelope drawing.
Solution Approach 2:
The patent dynamically adjusts the processing approach based on image characteristics. By using adaptive thresholding and connected domain analysis, the system can effectively identify blood flow spectra regardless of gain settings, maintaining reliability whether the gain is high (with complex background) or low (with simpler background but potentially lower brightness).
3Reliability
If manual marking is used to draw blood flow spectrum envelope to ensure accuracy, then the reliability of envelope drawing is improved, but automation is hindered and time consumption increases
Solution Approach 1:
The patent implements self-service by enabling the system to automatically draw blood flow spectrum envelopes without manual intervention. The algorithm automatically identifies block domains, extracts contour pixel points, and generates the envelope curve through coordinate connection, making the system self-sufficient and eliminating the need for manual marking while maintaining reliability through robust image processing and analysis.
Solution Approach 2:
The patent replaces the mechanical manual marking process with an automated computer-based image recognition and processing system. By substituting manual operations with algorithms that perform threshold processing, connected domain analysis, contour extraction, and coordinate-based envelope drawing, the system achieves both automation and reliability, eliminating time consumption associated with manual marking.
4Extent of automation
If computer image recognition technology is applied to draw blood flow spectrum envelope to achieve automation, then automation is improved, but the difficulty of detecting and measuring increases due to image complexity and noise
Solution Approach 1:
The patent applies preliminary image processing actions including Gaussian smoothing to reduce noise and threshold processing to enhance contrast between blood flow spectra and background. These preliminary steps simplify the image data before the main recognition process, significantly reducing the difficulty of detecting and measuring blood flow regions while maintaining automation.
Solution Approach 2:
The patent introduces intermediate processing steps and concepts, including binary image conversion, connected domain identification, and maximum domain extraction, as mediators between the raw complex ultrasound image and the final envelope drawing. These intermediary processes break down the complex detection task into manageable steps, reducing overall difficulty while achieving automation.
Data Source
AI summary
A generation method and device for a blood flow spectrum envelope includes: importing an image, then finding a connected domain; obtaining a contour map through the connected domain; marking coordinates of contour pixel points on the contour map; and connecting all the coordinates to form a blood flow spectrum envelope. A program stored in the storage medium or a processor of an ultrasound device has the steps of the generation method for the blood flow spectrum envelope. The blood flow spectrum envelope is automatically found; and when the image is continuously collected by ultrasound, the envelope is generated in real time according to the image collected by ultrasound in the disclosure to analyze a blood flow.