Medical Image Processing Using Adaptive Sub-Region Thresholding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Magnetic resonance imaging (MRI) images, such as T1, T2, and diffusion-weighted images, are not effectively interpreted by most users due to their complexity, making it difficult to distinguish and visualize abnormal tissues within medical images.
Innovation Solution
A system and method for processing medical images that identify target elements in a region of interest (ROI) by determining different thresholds for sub-regions based on average signal intensity and standard deviation, assigning color values to these elements, and generating a presentation that highlights abnormal tissue distribution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If MRI images are processed using conventional methods, then the imaging data is preserved in its original form, but the images are not sufficiently meaningful and poorly understood by users
Solution Approach 1:
The patent applies color coding to visual elements in medical images, where different colors represent different categories of elements (e.g., abnormal tissues, anatomical structures). This transforms the grayscale or single-channel MRI data into a multi-channel color representation that is more intuitive and easier to interpret for users, directly addressing the interpretability issue without requiring complex additional hardware
Solution Approach 2:
The system transforms the original imaging parameters (signal intensities, spatial coordinates) into new presentation parameters (color values, visual weights, categorical labels). This parameter transformation makes the data more meaningful to users while maintaining the underlying scientific accuracy, resolving the contradiction between preserving original data and improving interpretability
2Ease of operation
If the image processing system uses simple visualization methods, then the system is easy to operate, but it cannot effectively distinguish and visualize abnormal tissues
Solution Approach 1:
The patent segments the medical image into multiple categories of elements (abnormal tissues, normal tissues, anatomical landmarks) and assigns different visual properties to each segment. This segmentation enables precise identification of abnormal tissues while presenting the information in a simplified, easily interpretable visual format that does not require complex operational procedures from users
Solution Approach 2:
The system introduces an intermediary processing layer that automatically performs complex analysis (thresholding, classification, color assignment) between the raw MRI data and the final visual presentation. This intermediary handles the computational complexity internally while presenting simple, intuitive visual outputs to users, resolving the contradiction between operational simplicity and measurement precision
Data Source
AI summary
A method may include obtaining an image representing a region of interest (ROI) of an object. The ROI may include two or more sub-regions. The method may include determining an average value of quantitative indexes associated with elements in the image corresponding to a first region of the ROI. The method may include determining, for each of the two or more sub-regions of the ROI, a threshold based on the average value; identifying target elements in the image based on the thresholds of the two or more sub-regions. The method may include assigning a presentation value to each of at least some of the target elements based on the average value and the quantitative index of the each target element. The method may include generating a presentation of the image based on the presentation values.