Medical Image Evaluation Using Personalized Reference Ranges
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for evaluating medical image data often require reference image data from a healthy state, which is not available in most cases, limiting their effectiveness in assessing anatomical and functional changes in examination subjects.
Innovation Solution
A method that acquires a clinical marker specific to the examination subject, such as genetic or blood data, to determine a personalized normal value range for physiological parameters, allowing for the comparison of medical image data values with individually tailored reference values, thereby enhancing the sensitivity of medical image data evaluation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If reference image data from healthy state are used for evaluation, then measurement precision is improved, but availability of reference data deteriorates
Solution Approach 1:
The patent creates virtual reference image data by training a deep neural network on available medical image data from various subjects. The trained network generates synthetic reference images that mimic healthy state anatomy, replacing the need for actual reference images from the same subject. This copying approach maintains evaluation precision while solving the availability problem.
Solution Approach 2:
The patent transforms the evaluation approach by changing from direct image comparison to a learned parameter-based comparison. The deep neural network learns optimal parameters and features from training data, then uses these learned parameters to evaluate new subjects. This parameter transformation enables precision evaluation without requiring matching reference images.
2Ease of operation
If standard normal value ranges are used for comparison, then ease of operation is improved, but measurement precision deteriorates
Solution Approach 1:
The patent makes the normal value ranges dynamic rather than static. Instead of using fixed standard ranges, the system adapts the reference values based on individual subject characteristics extracted from their medical images. This dynamic adaptation maintains ease of operation while significantly improving measurement precision and detection sensitivity.
Solution Approach 2:
The patent performs preliminary analysis of each subject's medical image data to establish personalized reference ranges before conducting the actual evaluation. The deep neural network pre-processes subject-specific features and determines individualized normal ranges, which then serve as the basis for accurate pathology detection. This preliminary action bridges simplicity and precision.
Data Source
AI summary
In a method and computer for evaluating medical image data of an examination subject, a clinical marker of the examination subject is acquired that characterizes a status of the examination subject in relation to a physiological parameter, and a normal value range for the physiological parameter is ascertained that is matched to the status of the examination subject as a function of the clinical marker. Medical image data of the examination subject are acquired, and a value of the physiological parameter of the examination subject is determined using the medical image data. This value is compared with the normal value range matched to the status of the examination subject, and the result of the comparison is provided as an output.