Image Normalization via Chunk-Based Mapping for Digital Pathology
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital pathology methods face challenges in reproducibility and objectivity due to subjective human judgments in analyzing microscope images, leading to variations in diagnostic and prognostic assessments across different pathologists and imaging systems.
Innovation Solution
The method involves normalizing image pixel data by estimating a mapping transformation using low-resolution data to apply to high-resolution images, enabling objective comparisons and improving computational efficiency in automated diagnoses and prognostic assessments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If subjective human judgment is used in analyzing microscope images, then diagnostic assessments can be made using scientific knowledge and personal experience, but reproducibility deteriorates as different pathologists make different judgments based on identical images
Solution Approach 1:
The patent replaces the mechanical system of human visual inspection and subjective judgment with an automated image analysis system that processes microscope images through computational algorithms. This substitution eliminates inter-observer variability while maintaining diagnostic capability through objective, reproducible measurements of image features.
Solution Approach 2:
The patent introduces an intermediate processing layer between image acquisition and diagnostic interpretation, consisting of automated feature extraction and quantification algorithms. This intermediary system transforms subjective visual assessment into objective, measurable parameters that can be consistently analyzed across different cases and observers.
2Measurement precision
If multiple microscope images are analyzed from a single patient's tissue sample, then comprehensive diagnostic assessment is achieved, but the problem of reproducibility compounds as pathologists may focus on different areas of different images
Solution Approach 1:
The patent segments the analysis process into distinct computational stages: image preprocessing, feature extraction, quantification, and interpretation. By dividing the comprehensive analysis of multiple images into systematic segments, the system ensures that all relevant areas are evaluated consistently without relying on individual pathologist focus or judgment variability.
Solution Approach 2:
The patent creates a universal analysis framework that processes all microscope images from a patient sample through the same standardized algorithms and criteria. This multi-functional system handles multiple images, regions, and feature types uniformly, ensuring reproducible results regardless of which images or areas are analyzed.
3Productivity
If automated processing methods are implemented for digital pathology slides, then productivity increases through high-throughput analysis, but the ability to provide objective comparisons between different tissue samples acquired using different systems deteriorates due to system-specific variations
Solution Approach 1:
The patent applies parameter transformation and normalization techniques to standardize image data from different imaging systems. By converting system-specific parameters into standardized, comparable metrics, the system maintains inter-system comparability while enabling high-throughput automated processing of diverse tissue samples.
Solution Approach 2:
The patent creates an equipotential processing environment where images from different systems are transformed into a common reference frame with standardized characteristics. This level playing field ensures that all samples are processed and compared under equivalent conditions, eliminating system-specific biases while maintaining processing efficiency.
4Measurement precision
If high-resolution image data is processed directly, then diagnostic accuracy is maintained, but computational efficiency deteriorates due to the large volumes of data requiring analysis
Solution Approach 1:
The patent segments the large-volume high-resolution image data into manageable processing units or regions of interest. By dividing the comprehensive image set into smaller segments, the system can process and analyze critical features efficiently while maintaining diagnostic accuracy through targeted examination of relevant areas.
Solution Approach 2:
The patent extracts and prioritizes diagnostically relevant features and regions from the full high-resolution image data. By taking out only the essential information needed for accurate diagnosis, the system reduces computational load and processing time while preserving diagnostic accuracy through focused analysis of key features.
Data Source
AI summary
A computer implemented image processing for normalizing images is disclosed. The method comprises calculating, for each chunk of an image, a combination of the two largest contributions to the variations of color of the chunk that approximates a representation of three additive color component data elements of the each image chunk, thereby to provide a plurality of approximated image chunks. The approximated image chunks are used to determine (i) a first item of approximated image chunk data, vMin, that corresponds to a first one of the two largest contributions, and (ii) a second item of the approximated image chunk data, vMax, that corresponds to a second one of the two largest contributions, wherein the first item of approximated image chunk data and the second item of approximated image chunk data together provide a mapping transformation and for each image pixel, image pixel data is obtained comprising three additive color component data elements of the each image pixel, and transformed using the mapping transformation to calculate normalized image pixel data for the each image pixel.