Medical Image Fusion Using Parameter Multiplication for Lesion Contrast
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing medical image fusion methods, both conventional superposition and deep learning-based, fail to accurately identify lesions due to low contrast and resolution, particularly in cavity organs, and are dependent on insufficient training data, leading to missed or misdiagnoses.
Innovation Solution
A method involving the multiplication of parameter values from different medical images, such as CT and PET/SPECT, to generate synthetic images with enhanced contrast and resolution, utilizing parameter values like linear attenuation coefficients and standardized uptake values to amplify differences between normal and lesion tissues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional superposition fusion method is used to combine PET/SPECT and CT images, then the fusion image can be generated, but the contrast and resolution remain low making lesion identification inaccurate
Solution Approach 1:
The patent transforms the fusion approach by changing from simple superposition to multiplication operation in the attenuation coefficient space. This parameter transformation enables the functional image parameters (standardized uptake values) to modulate the anatomical image parameters (attenuation coefficients), thereby enhancing lesion contrast while maintaining quantitative accuracy.
Solution Approach 2:
The patent introduces attenuation coefficients as an intermediary medium to bridge PET/SPECT functional data and CT anatomical data. By performing multiplication in this intermediate representation space rather than direct image space, the method achieves superior fusion quality that preserves both functional and anatomical information with enhanced lesion visibility.
2Extent of automation
If deep learning-based fusion method is used, then automated fusion can be achieved, but the method is highly dependent on training data quantity and quality affecting performance
Solution Approach 1:
The patent creates a self-sufficient fusion method that does not require external training data or pre-trained models. The multiplication-based fusion approach is inherently robust and produces reliable results without needing to learn from large datasets, thereby achieving both automation and reliability simultaneously.
Solution Approach 2:
The patent replaces the complex mechanical learning system (deep neural networks requiring extensive training) with a simpler mathematical operation (multiplication in attenuation coefficient space). This substitution maintains automation while eliminating the reliability issues associated with data-dependent learning methods.
3Productivity
If deep learning-based fusion is applied, then fusion processing can be performed, but the resolution of fused images remains relatively low affecting small lesion detection
Solution Approach 1:
The patent changes the operational domain from pixel intensity space to attenuation coefficient space, where multiplication operations naturally preserve and enhance edge information. This parameter transformation results in fused images with superior resolution that enables accurate detection of small lesions while maintaining efficient processing.
4Loss of information
If standardized uptake values from PET/SPECT are used directly, then functional information is preserved, but normal tissues with high uptake obscure small lesions leading to missed diagnosis
Solution Approach 1:
The patent applies local quality enhancement by using multiplication to selectively amplify functional information in regions where it is most useful. The attenuation coefficient multiplication naturally emphasizes areas with both anatomical structure and functional activity, while suppressing regions where high normal uptake obscures pathology, thereby achieving local optimization of lesion visibility.
Data Source
AI summary
Embodiments of the present application provide a medical image synthesis device and method. According to an embodiment, a method includes acquiring a first medical image and a second medical image and registering the first medical image with the second medical image. The method includes determining a first parameter value at each pixel location on the registered first medical image and a second parameter value at each pixel location on the second medical image. The method includes multiplying the first parameter value with the second parameter value at the same pixel location on the registered first medical image and the second medical image and generating synthetic image data based on the multiplication result.