Gleason scoring via co-registered tissue image segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual Gleason grading of prostate tissue is prone to inconsistencies and missed areas, leading to inaccurate Gleason scores due to the labor-intensive and subjective evaluation of stained tissue samples by pathologists.
Innovation Solution
A method involving digital image analysis of co-registered slides stained with different biomarkers, where image objects are generated and analyzed to determine histopathological scores, including Gleason scores, by identifying and classifying patterns in prostate tissue using features like asymmetry, roundness, and separation from basal epithelial cells, thereby reducing human error and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual Gleason grading is performed by pathologists visually evaluating stained tissue samples, then diagnostic capability is achieved, but grading accuracy and consistency deteriorate due to human error and subjectivity
Solution Approach 1:
The patent replaces the manual mechanical evaluation process with an automated digital image analysis system. The system uses computer algorithms to objectively evaluate tissue architecture patterns, glandular structures, and spatial relationships in digital images of stained tissue sections, eliminating human subjectivity and fatigue-related errors while maintaining diagnostic capability.
Solution Approach 2:
The patent creates digital copies of physical tissue slides through high-resolution scanning. These digital replicas can be analyzed repeatedly without degradation, allowing multiple pathologists or algorithms to evaluate the same tissue sample and enabling automated measurement of architectural patterns that would be difficult to assess consistently on physical slides.
2Measurement precision
If pathologists manually inspect all cancerous areas on highly magnified tissue images, then comprehensive evaluation is achieved, but time consumption and labor intensity increase significantly
Solution Approach 1:
The patent segments the tissue image into distinct architectural patterns and glandular structures using image processing algorithms. The system automatically identifies and separates different tissue regions, cancerous areas, and glandular units, enabling comprehensive evaluation of all cancerous areas without requiring manual inspection of each individual structure.
Solution Approach 2:
The patent introduces digital image processing software as an intermediary between the physical tissue sample and the pathologist's evaluation. The software pre-processes the images by enhancing contrast, segmenting structures, and highlighting relevant features, thereby reducing the time pathologists need to spend on manual inspection while ensuring comprehensive coverage of all cancerous areas.
3Measurement precision
If multiple tissue slices are stained and analyzed, then diagnostic information is improved, but processing complexity and resource requirements increase
Solution Approach 1:
The patent merges multiple digital images of differently stained tissue slices into a co-registered composite image. The system aligns and overlays images from different staining protocols (e.g., H&E, immunohistochemistry) to create a unified view that preserves spatial relationships and enables simultaneous evaluation of multiple tissue characteristics in a single analysis framework.
Solution Approach 2:
The patent develops a universal digital image analysis platform that can handle multiple staining types and tissue types through a single system. The software is designed to process various image formats and staining protocols using common algorithms, thereby reducing the need for separate specialized systems while maintaining high diagnostic information quality.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
An improved histopathological score is obtained by generating image objects from images of tissue containing stained epithelial cells. First objects are generated that correspond to basal cells stained with a first stain, such as p63. Second objects are generated that correspond to luminal cells stained with a second stain, such as CK18. If the same tissue is not stained with both stains, then the images of differently stained tissue are co-registered. Third objects are defined to include only those second objects that have more than a minimum separation from any first object. A scoring region includes the third objects, and the histopathological score is determined based on tissue that falls within the scoring region. For example, a Gleason score of prostate tissue is determined by classifying tissue patterns in the scoring region. Alternatively, a Gleason pattern is assigned by counting the number of third objects that possess a predetermined form.