Ultrasound Image Feed Analysis for Continuous Tissue Highlighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing workflows for integrating machine learning models in ultrasound examinations, particularly for breast US examinations, face challenges due to the inherent coupling of image acquisition and user interpretation, leading to inconsistent and distracting detection marks that hinder effective clinical performance.
Innovation Solution
A method and apparatus for analyzing ultrasound image feeds by detecting a target tissue above a visibility threshold and continuously highlighting it across consecutive images, ensuring consistent tracking of the target tissue even if visibility fluctuates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ML models are used to detect target tissue in ultrasound images, then detection accuracy is improved, but detection marks may flash or disappear when visibility fluctuates, causing distraction and reducing reliability
Solution Approach 1:
The system determines visibility thresholds for target tissue before actual detection occurs. By pre-establishing these thresholds based on image characteristics and tissue properties, the system ensures that detection marks are only displayed when visibility is sufficient, preventing flashing or disappearing marks that would distract the clinician and reduce reliability.
2Loss of information
If detection marks are displayed for all detected tissues, then information completeness is improved, but too many marks may confuse or distract the user, reducing ease of operation
Solution Approach 1:
The system applies different visibility thresholds to different regions or types of target tissue within the ultrasound image. By adjusting the visibility criteria locally based on tissue characteristics, image quality, and clinical context, the system displays only the most relevant detection marks, avoiding information overload while maintaining completeness of important findings.
3Productivity
If ML models are integrated into live ultrasound workflow, then productivity is improved, but the coupling of image acquisition and interpretation creates complexity in implementing consistent detection across consecutive images
Solution Approach 1:
The system pre-determines visibility thresholds and detection parameters before the live ultrasound examination begins. This preliminary configuration allows the ML model to operate consistently across consecutive images without requiring complex real-time adjustments, simplifying workflow integration while maintaining high productivity through automated detection throughout the examination.
Data Source
AI summary
According to an aspect, there is provided a computer implemented method of analysing an ultrasound, US, image feed as part of an ultrasound examination. The method comprises: detecting a target tissue in a first image of the US image feed, sending a message to a display to cause the display to highlight the target tissue in the first image on the display, if the target tissue has a visibility above a first visibility threshold, and causing the display to highlight the target tissue in each consecutive US image subsequent to the first image in which the target tissue is visible in the US image feed.