Digital Pathology Image Processing via Sub-Region Transform
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Digital pathology systems face performance issues due to large digital image sizes, requiring intensive computation and storage, and struggle with real-time image registration and display of multi-modal images with varying pixel intensity distributions and non-uniform staining procedures.
Innovation Solution
Applying image transformations only to high-resolution sub-regions on demand, using metrics for optimal transform function determination, and employing multi-resolution image data structures for efficient image registration and display, allowing real-time performance and improved quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image processing algorithms are applied to entire large digital images, then image quality enhancement is achieved, but computational intensity and storage requirements increase significantly
Solution Approach 1:
The patent divides the large digital image into multiple smaller sub-regions or tiles. Instead of processing the entire image at once, the system processes only the selected sub-regions, reducing computational intensity and storage requirements while maintaining image quality enhancement for the areas of interest.
Solution Approach 2:
The patent extracts only the necessary sub-regions from the large digital image for processing. By taking out only the relevant portions rather than processing the complete image, the system reduces computational overhead and storage needs while preserving the quality enhancement benefit for the extracted regions.
2Productivity
If down sampling is applied to reduce image size, then real-time performance is achieved, but image quality becomes insufficient
Solution Approach 1:
The patent applies different processing qualities to different regions. High-resolution processing is applied locally to selected sub-regions where detailed analysis is needed, while lower resolution is used for the remaining areas. This allows real-time performance for the processed regions while maintaining sufficient image quality where required.
Solution Approach 2:
The patent dynamically adjusts the processing resolution based on user interaction and analysis needs. The system can switch between different resolution levels and processing intensities in real-time, allowing high-quality processing when needed while maintaining overall real-time performance through adaptive processing strategies.
3Reliability
If image registration is applied to multi-modal images with varying pixel intensity distributions, then spatial alignment is achieved, but computational time increases
Solution Approach 1:
The patent segments the image registration process into multiple stages, processing different regions or features separately. By dividing the registration task into smaller sub-problems, the system achieves accurate spatial alignment for multi-modal images while reducing overall computational time through parallel processing and selective registration of critical regions.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
The invention pertains to the field of image processing in digital pathology. It notably proposes a method for processing a first digital image, representing a sample in a region, and which image has been acquired from the sample by means of a microscopic imaging system (1) and is stored in a multi-resolution image data structure (80), comprising the steps of: - retrieving (104) a sub-region of the first digital image at a first resolution, -executing (105) a transform function on the retrieved sub-region, the transform function modifying a content of the sub-region according to at least one metric derived from a second resolution representation of the first digital image.