Ultrasound Imaging Device with Deep-Learning Image Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing ultrasound imaging devices rely on subjective and qualitative criteria for selecting input images for diagnostic algorithms, leading to variability in results and reduced accuracy due to user-dependent factors and anatomical view discrepancies.
Innovation Solution
An ultrasound imaging device that quantitatively determines the suitability of ultrasound images for diagnostic algorithms by using pre-trained deep neural networks and objective criteria, allowing for the selection of optimal images based on anatomical views, quality, and scan parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If subjective and qualitative criteria are used for selecting input images, then ease of operation is improved, but measurement precision and reliability deteriorate due to user-dependent variability
Solution Approach 1:
The patent replaces the manual, subjective image selection process with an automated deep learning-based system. The examiner's subjective judgment is substituted by an automated algorithm that objectively evaluates multiple ultrasound images and selects the optimal input image for the diagnostic algorithm, thereby eliminating user-dependent variability while maintaining operational simplicity
Solution Approach 2:
The patent introduces an intermediate automated image selection system between the ultrasound image acquisition and the diagnostic algorithm. This intermediary component evaluates multiple images using deep learning and selects the most suitable one, preventing the direct transmission of subjective selection errors to the diagnostic process
2Measurement precision
If multiple ultrasound images are evaluated using deep learning, then measurement precision is improved, but device complexity and processing time increase
Solution Approach 1:
The patent performs preliminary evaluation of multiple ultrasound images using deep learning before the actual diagnostic algorithm execution. By pre-assessing image quality, anatomical view suitability, and scan condition appropriateness, the system prepares the optimal input image in advance, ensuring high measurement precision without adding complexity to the core diagnostic function
Solution Approach 2:
The patent divides the image evaluation process into distinct segments: quality assessment, anatomical view verification, and scan condition checking. Each segment is handled by specific deep learning models trained for particular tasks, allowing the system to achieve high precision through specialized sub-functions rather than a single complex system
3Reliability
If automated image selection is implemented, then reliability is improved by eliminating user variability, but device complexity increases due to additional processing requirements
Solution Approach 1:
The patent designs the automated image selection system to perform multiple functions simultaneously: quality assessment, anatomical view verification, and optimal image selection for the diagnostic algorithm. This multi-functional approach achieves high reliability through comprehensive evaluation while avoiding the complexity of separate dedicated systems for each function
Solution Approach 2:
The system enables the ultrasound imaging device to automatically evaluate and select optimal images without requiring external intervention or complex manual procedures. The deep learning models autonomously assess image suitability and make selection decisions, improving reliability by eliminating human variability while keeping the system architecture relatively simple through self-contained automation
Data Source
AI summary
Provided is an ultrasound imaging device and an operation method thereof. An embodiment of the present disclosure provides an ultrasound imaging device comprising: an ultrasound probe; a display unit; a memory for storing at least one instruction; and a processor for executing the at least one instruction stored in the memory, wherein the processor: image-processes an echo signal to acquire multiple ultrasound images; calculates suitability indicating, as a numerical value, whether the multiple acquired ultrasound images are suitable as input images for a diagnosis algorithm for diagnosing a lesion; displays the calculated suitability on the display unit; and determines an input ultrasound image, which is to be input into the diagnosis algorithm, among the multiple ultrasound images based on the suitability.