Digital Pathology Slide ROI Detection for Efficient Scanning

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing digital pathology slide scanning methods are manpower and resource intensive, inefficient due to scanning of blank and unrelated regions, and occupy large space.

Innovation Solution

A deep learning-based method and system for identifying target regions on digital pathology slides using image contour feature extraction, segmentation, and classification models to separate effective pathological, broken contaminant, and slide information regions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If comprehensive scanning is performed on the entire digital pathology slide, then no region is missed, but scanning time and resource consumption increase significantly

Engineering Contradiction:

Improvecompleteness of scanningVSAvoidscanning time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent segments the pathology slide into multiple regions of interest (ROIs) based on color features and texture analysis. Instead of scanning the entire slide uniformly, the system divides it into effective regions (containing pathological information) and ineffective regions (blank or unrelated areas), then applies different scanning strategies to each segment, thereby reducing overall scanning time while maintaining completeness of important regions

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent implements partial scanning by identifying and prioritizing only the necessary regions for diagnosis. The system performs comprehensive scanning on effective pathological regions while using simplified or skipped scanning on blank and unrelated regions, achieving a balance between completeness and efficiency through selective application of scanning depth

Inventive Principle:

Principle #16Partial or excessive action

2Measurement precision

If manual selection of to-be-scanned regions is performed by naked eyes, then scanning focus is improved, but manpower consumption and subjectivity increase

Engineering Contradiction:

Improveregion selection accuracyVSAvoidautomation level

Core Design Contradiction:

Measurement precisionVSExtent of automation

Solution Approach 1:

The patent replaces the manual visual inspection mechanism with an automated computer-based image analysis system. The system uses color feature extraction, texture analysis, and machine learning algorithms to automatically identify and classify regions, substituting human expert judgment with objective computational methods that maintain high accuracy while eliminating subjectivity and labor consumption

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent enables the scanning system to automatically identify and prioritize regions of interest without requiring manual intervention. The image analysis algorithms self-assess the importance of different slide regions based on predefined criteria (color, texture, structural features), allowing the system to autonomously determine scanning priorities and execute the scanning plan without human input

Inventive Principle:

Principle #25Self-service

3Reliability

If comprehensive scanning of the entire slide is performed, then all regions are captured, but storage space requirements increase significantly

Engineering Contradiction:

Improvecompleteness of data captureVSAvoidstorage space

Core Design Contradiction:

ReliabilityVSVolume of stationary object

Solution Approach 1:

The patent extracts and stores only the essential pathological information from the slide by identifying effective regions through image analysis. Instead of storing complete high-resolution images of the entire slide, the system extracts key regional features, generates representative images of identified ROIs, and stores only these extracted elements, significantly reducing storage requirements while preserving diagnostic information

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies different quality levels and storage strategies to different regions of the slide. Effective pathological regions are captured and stored with high quality and detail, while blank and unrelated regions are either skipped or stored with reduced quality, creating a non-uniform storage approach that optimizes space usage while maintaining the quality of diagnostically important areas

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS12579828B2Method and system for identifying target region of digital pathology slide, device, and medium

Publication Date: 2026.03.17 GUANGZHOU KINGMED CENTER FOR CLINICAL LABORATORY CO LTD

AI summary

A method for identifying a target region of a digital pathology slide, including: obtaining a scanned image of a pathology slide; inputting the scanned image of the pathology slide into a preset deep learning-based identification model; extracting a contour feature of the scanned image of the pathology slide by using an image contour feature extraction submodel, to obtain a contour image; segmenting the contour image by using an image segmentation submodel to obtain a plurality of sub-contour images; separately performing classification and identification on the plurality of sub-contour images by using an image classification submodel, to obtain a region category corresponding to each sub-contour image; and determining a target region image based on the region category of each sub-contour image. In addition, a system for identifying a target region of a digital pathology slide, a device, and a medium are further proposed.