Automated Prioritization of Digital Pathology Slides Using Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is no standardized or efficient way to prioritize the review of images of tissue specimens for pathology patient cases, leading to inefficiencies in the digital pathology workflow.
Innovation Solution
A computer-implemented method and system that uses a machine learning system to compute a prioritization value for electronic images of tissue specimens, based on training images labeled with characteristics such as slide morphology, diagnostic value, pathologist review outcome, and analytic difficulty, to automatically prioritize the processing and review of pathology slides.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If manual review and prioritization of pathology slides is performed, then pathologists can review slides in order of importance, but the process is time-consuming and lacks standardization
Solution Approach 1:
The system enables automated self-service prioritization of pathology slides through machine learning algorithms that automatically analyze slide images, extract features, and rank slides by diagnostic importance without requiring manual pathologist intervention for prioritization
Solution Approach 2:
The patent replaces the manual mechanical process of pathologist-based slide prioritization with an automated computational system that uses machine learning models to perform feature extraction, analysis, and ranking, substituting human manual review with algorithmic processing
2Reliability
If all pathology slides are reviewed thoroughly, then diagnostic accuracy is maintained, but the overall processing time increases
Solution Approach 1:
The system segments the slide review process into two distinct phases: automated prioritization phase that ranks all slides by importance, and selective detailed review phase that focuses pathologist attention on high-priority slides, allowing thorough review of critical cases while maintaining overall productivity
Solution Approach 2:
The system applies partial action by performing comprehensive automated feature extraction and analysis on all slides to determine prioritization, while human pathologist review is applied selectively only to high-priority slides rather than all slides, optimizing the balance between thoroughness and efficiency
3Productivity
If insufficient slides are reviewed and processed, then processing speed increases, but diagnostic quality may be compromised
Solution Approach 1:
The system performs preliminary automated quality assessment and prioritization of slides before they reach the pathologist for final review, identifying and flagging insufficient or low-quality slides in advance, allowing pathologists to focus on adequate slides while maintaining both speed and quality standards
Data Source
AI summary
Systems and methods are disclosed for processing digital pathology images, prioritizing the digital pathology images, and outputting a sequence of the digital pathology images based on the prioritization. The prioritization may be determined by a machine learning model trained to determine prioritization values based on various criteria. For example, the machine learning may generate biomarker expression information and determine prioritization values based on the generated information.