Ultrasound Tissue Segmentation Using RF Phase Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current ultrasound image analysis techniques face challenges in accurately differentiating between tissues with similar acousto-mechanical properties, leading to difficulties in precise boundary delineation and increased false positive results, especially when using conventional amplitude-only greyscale images, which lack the detailed information present in raw radio frequency (RF) data.
Innovation Solution
A method involving a computing device that receives an ultrasound image, assigns labels to each pixel using a segmentation machine-learning model, and classifies the image into diagnostic classes based on a classification machine-learning model, utilizing both grey ultrasound and RF data to enhance tissue differentiation and reduce false positives by leveraging the additional information in RF data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional amplitude-only greyscale images are used for ultrasound analysis, then the imaging process remains simple and fast, but tissue differentiation accuracy deteriorates when adjacent tissues have similar acousto-mechanical properties
Solution Approach 1:
The patent transitions from analyzing only amplitude information (1D) to utilizing both amplitude and phase information (2D), effectively adding a dimensional aspect to the ultrasound data. This allows differentiation of tissues with similar acousto-mechanical properties by exploiting phase variations that are invisible in conventional greyscale images.
Solution Approach 2:
The patent changes the analytical parameters from solely amplitude-based to include both amplitude and phase parameters. By analyzing RF data in the frequency domain and extracting phase information, the system achieves improved tissue characterization without requiring complex hardware modifications.
2Measurement precision
If whole-image classification is used for ultrasound diagnosis, then the analysis process is simplified, but diagnostic accuracy deteriorates due to false positive results
Solution Approach 1:
The patent divides the ultrasound image into multiple local regions and performs classification on each region independently rather than treating the entire image as a single unit. This segmentation approach reduces false positives by capturing local tissue characteristics that would be averaged out in whole-image classification, while the modular nature of the approach keeps computational complexity manageable.
Solution Approach 2:
The patent applies different classification criteria and parameters to different local regions of the ultrasound image based on their specific characteristics. Each region is analyzed with attention to its local acousto-mechanical properties, allowing for more nuanced and accurate diagnostic assessment compared to uniform whole-image analysis.
3Measurement precision
If RF data analysis is implemented to capture frequency and phase information, then tissue characterization accuracy improves, but computational requirements and processing complexity increase
Solution Approach 1:
The patent extracts only the essential features from the full RF data - specifically amplitude and phase information at key frequency points - rather than processing the entire RF signal spectrum. This extraction approach maintains high tissue characterization accuracy while significantly reducing computational energy requirements compared to full-spectrum analysis.
Solution Approach 2:
The patent applies partial action by focusing computational resources on analyzing only the most diagnostically relevant frequency and phase components of the RF data, rather than processing all available information equally. This selective approach achieves effective tissue characterization with reduced computational energy consumption.
Data Source
AI summary
Provided are methods including the steps of receiving, with at least one computing device, an image of a portion of a subject, assigning: with the at least one computing device and based on a machine-learning model, a label to one or more pixels of the image to generate a diagnostically segmented image: and classifying, with the at least one computing device and based on a machine-learning model, the diagnostically segmented image and the one or more pixels into at least one class to generate a classified image, wherein the classified image includes a classification label indicating a clinical assessment of the portion of the subject and wherein the one or more pixels include a clinical label indicating a diagnosis of a portion of a subject contained within each pixel, based on the diagnostically segmented image having labels assigned to each pixel of the segmented image.