AI Microscope Image Overlay for Faster Feature Verification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current AI-based image recognition in microscopic images suffers from low accuracy and requires manual verification of all results, increasing workload due to complexity in feature recognition and low image resolution.

Innovation Solution

An image processing method using AI algorithms to recognize features in microscopic images, followed by superimposing the recognition results onto the image through augmented reality, allowing for preliminary filtering and verification of only those sections identified as potentially containing the feature, thereby reducing unnecessary verification workload.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If AI algorithms are used for feature recognition in microscopic images, then recognition speed is improved, but recognition accuracy deteriorates due to complexity in feature recognition and low image resolution

Engineering Contradiction:
Improverecognition speedVSAvoidrecognition accuracy
Core Design Contradiction:
SpeedVSMeasurement precision

Solution Approach 1:

The patent segments the image processing workflow into two distinct stages: (1) AI-based preliminary screening to identify candidate regions, and (2) manual verification focused only on those candidates. This segmentation allows the system to leverage AI's speed advantage while minimizing its accuracy limitations by restricting its role to initial filtering rather than final determination.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies preliminary action by using AI algorithms to perform preliminary screening and identification of candidate regions before manual verification. This preliminary action filters out obvious non-matching cases, so that subsequent manual verification only needs to focus on a small subset of promising candidates, thereby improving overall efficiency without compromising accuracy.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If manual verification is performed on all AI recognition results, then recognition accuracy is improved, but workload increases

Engineering Contradiction:
Improverecognition accuracyVSAvoidworkload
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent applies partial action by performing manual verification only on a subset of AI-identified candidate regions rather than on all AI recognition results. This partial verification approach maintains high accuracy for critical cases while significantly reducing the overall manual workload compared to verifying every AI output.

Inventive Principle:
Principle #16Partial or excessive action

3Adaptability or versatility

If feature recognition complexity increases, then recognition capability is improved, but processing time increases

Engineering Contradiction:
Improverecognition capabilityVSAvoidprocessing time
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent segments the processing into two phases with different complexity requirements: Phase 1 uses AI for rapid preliminary screening with moderate complexity, and Phase 2 uses manual verification with high complexity only on selected candidates. This segmentation allows the system to achieve high recognition capability where needed while minimizing time loss through the use of simpler, faster AI processing for the bulk of cases.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3940585B1Image processing method based on artificial intelligence, microscope, system and medium
Publication Date: 2026.04.22 TENCENT TECHNOLOGY (SHENZHEN) CO LTD
  • EP3940585B1 patent drawingFigure 1~2A
  • EP3940585B1 patent drawingFigure 2B~2C
  • EP3940585B1 patent drawingFigure 3A~3B

AI summary

The embodiment of the present disclosure provides an image processing method based on artificial intelligence, a microscope, a system and a medium. The image processing method based on artificial intelligence comprises: obtaining a feature recognition result of an image, wherein the feature recognition result is obtained by conducting image processing on the image to recognize features in the image, the image is obtained by conducting image collection on a slice; determining an imaging area in the visual field of an ocular lens for imaging the slice by a microscope; determining an image area corresponding to the imaging area in the image, obtaining a feature recognition result of the image area as a to-be-displayed recognition result; and displaying the to-be-displayed recognition result on the imaging area in an overlapping manner.