Histology Image Risk Stratification Using Excluded Intermediate Grades
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current histologic grade assessment in cancer, particularly for breast cancer progression risk, suffers from significant inter-observer variability and errors in classification, especially in the intermediate risk group (NHG 2), leading to suboptimal treatment decisions.
Innovation Solution
A method using a trained neural network to predict cancer progression risk by processing digital images of histological samples, excluding images of intermediate risk during training, and dividing images into sub-areas for improved classification, allowing for clearer differentiation between low and high risk groups.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual histologic grade assessment by pathologists is used, then treatment selection can be made, but inter-observer variability leads to errors in risk classification
Solution Approach 1:
The patent replaces the manual mechanical assessment process by pathologists with an automated digital image analysis system using machine learning algorithms. The system processes digital histology images through trained neural networks to objectively determine histologic grade and cancer progression risk, eliminating human inter-observer variability while maintaining or improving assessment accuracy.
2Adaptability or versatility
If the intermediate risk group (NHG 2) is included in training data, then the model covers all patient populations, but the heterogeneity of NHG 2 reduces classification precision
Solution Approach 1:
The patent extracts and separates the intermediate risk group (NHG 2) from the training dataset, focusing the neural network training exclusively on clear-cut low risk (NHG 1) and high risk (NHG 3) cases. This extraction allows the model to learn distinct morphological features of well-defined groups first, then apply this knowledge to classify heterogeneous intermediate cases with improved precision by comparing them against the learned extremes.
3Reliability
If digital image analysis with neural networks is implemented, then inter-observer variability is reduced, but complex data processing and model training are required
Solution Approach 1:
The patent performs preliminary actions by pre-training the neural network model on large datasets of labeled histology images before deployment. The model is pre-trained to recognize morphological features associated with different risk groups, and this preliminary training phase captures the complex patterns that will enable consistent classification during actual use, reducing the complexity burden during operational phases.
4Adaptability or versatility
If the Nottingham grading system is used with three grade groups, then comprehensive risk stratification is achieved, but the intermediate group exhibits large variability in morphological patterns
Solution Approach 1:
The patent segments the classification task into two phases: first training on well-defined low and high risk groups to learn extreme morphological patterns, then using this segmented knowledge to evaluate intermediate cases. This segmentation approach allows the system to handle the morphological variability in NHG 2 by comparing intermediate cases against the learned extremes rather than attempting to classify them directly during training.
Data Source
AI summary
There is provided a method comprisingdetermining cancer progression risk for a cancer patient by providing a digital image to a trained neural network and allowing the trained neural network to predict cancer progression risk for the patient based on that image, where the neural network has been trained byreceiving a training dataset comprising digital images of histology samples from cancer patients where each histology sample is associated in the dataset with one histology grade score selected from a set comprising three histology grade scores: a first histology grade score indicating low risk for progression of the cancer disease, a second histology grade score indicating intermediate risk for progression of the cancer disease and a third histology grade score indicating high risk for progression of the cancer disease,using the digital images of histology samples associated with the first and third histology grade scores, while ignoring digital images associated with the second histology grade score, to train a neural network for determining the cancer progression risk for a patient.