Mobile Pathological Image Labeling via FOV Segmentation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current methods for pathologically labeling medical images face challenges in ensuring labeling quality and efficiency, particularly when using artificial intelligence, as they require large capacity computers and are limited by working location, making it difficult to generate a huge number of high-quality learning samples.

Innovation Solution

A method and device that allow users to perform pathological labeling on a mobile terminal by dividing original scanning images into smaller FOV images, calculating their pathological index, and selecting the highest index images for labeling, which reduces workload and fatigue, enabling labeling anywhere and providing a system for quality control and consensus building among users.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If whole and integral labeling is performed on large capacity computers, then labeling quality is maintained, but working location is limited and labeling efficiency is reduced

Engineering Contradiction:

Improveworking location flexibilityVSAvoidlabeling efficiency

Core Design Contradiction:

Ease of operationVSProductivity

Solution Approach 1:

The patent divides the original scanning image into multiple FOV (field of view) images, allowing pathologists to label only the relevant regions instead of the entire slide. This segmentation enables labeling on mobile terminals with smaller screens while maintaining efficiency, as pathologists can quickly review and label only the areas of interest without being constrained by large computer requirements.

Inventive Principle:

Principle #1Segmentation

2Reliability

If whole and integral labeling is performed, then labeling completeness is ensured, but workload increases and fatigue increases

Engineering Contradiction:

Improvelabeling completenessVSAvoidlabeling time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent calculates a pathological index for each FOV image to identify and prioritize regions with the highest pathological significance. Pathologists then focus their labeling efforts on these high-priority regions rather than uniformly reviewing the entire slide. This local quality approach ensures that labeling completeness is maintained for the most critical areas while significantly reducing overall workload and labeling time.

Inventive Principle:

Principle #3Local quality

3Productivity

If AI labeling is used without systematic quality control, then labeling speed increases, but labeling accuracy and reliability decrease

Engineering Contradiction:

Improvelabeling speedVSAvoidlabeling accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent implements a systematic quality control mechanism that provides feedback to pathologists during the labeling process. The system calculates pathological indices, identifies high-priority regions, and enables pathologists to review and correct AI-generated labels. This feedback loop maintains high labeling speed while ensuring accuracy, as pathologists can quickly verify and correct errors without significantly slowing down the overall labeling process.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS11094411B2Methods and devices for pathologically labeling medical images, methods and devices for issuing reports based on medical images, and computer-readable storage media

Publication Date: 2021.08.17 GUANGZHOU KINGMED CENTER FOR CLINICAL LABORATORY CO LTD

AI summary

Disclosed herein are a method and a device for pathologically labeling medical images, which is capable of effectively solving the problem associated with limitations on working location of artificial labeling, enhancing labeling efficiency, and providing a huge number of learning samples with high quality with artificial intelligence.