Automated Area Selection in Charged Particle Microscopy

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Charged particle microscopy (CPM) systems require extensive manual intervention for area-of-interest selection, leading to stagnant throughput and inefficiency, despite advances in technology.

Innovation Solution

Implementing a machine-learning model trained on selection data to automatically determine areas for microscopy imaging, using a fully convolutional neural network for improved accuracy and efficiency, and integrating it into the CPM system for automated area selection.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If manual area-of-interest selection is used in CPM systems, then operator control and decision-making are maintained, but system throughput and operational efficiency deteriorate due to extensive manual intervention

Engineering Contradiction:
Improvemanual controlVSAvoidsystem throughput
Core Design Contradiction:
Ease of operationVSProductivity

Solution Approach 1:

The system performs area selection automatically using machine learning models without requiring manual operator intervention. The ML model analyzes microscopy images and autonomously identifies areas of interest, allowing the system to serve itself rather than relying on human operators for each selection decision.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces the manual mechanical process of area selection with an automated computational system. Machine learning algorithms process images and make selection decisions, substituting the mechanical human-in-the-loop approach with an automated intelligent system that maintains accuracy while dramatically improving throughput.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Productivity

If automated machine-learning models are implemented for area selection, then system throughput and efficiency improve, but system complexity and computational requirements increase

Engineering Contradiction:
Improvesystem throughputVSAvoidsystem complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The system performs preliminary training of machine learning models using labeled data before deployment. This preliminary action creates a trained model that can then be reused for automated area selection, reducing the complexity of real-time operations while maintaining high throughput capabilities.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces machine learning models as intermediary components between image acquisition and area selection. These models act as mediators that process images and generate selection decisions, managing the complexity by encapsulating the intelligent decision-making process in separate modular components rather than embedding it throughout the entire system.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Measurement precision

If extensive manual intervention is required for area selection, then selection accuracy can be maintained through expert judgment, but time consumption and operational inefficiency increase

Engineering Contradiction:
Improveselection accuracyVSAvoidtime consumption
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces manual expert judgment with machine learning models that have been trained on labeled data representing expert selections. This substitution maintains selection accuracy by capturing expert knowledge in the trained model while eliminating the time consumption associated with manual intervention for each selection task.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system creates copies of expert knowledge through labeled training data and trained machine learning models. Instead of requiring actual expert operators for each selection, the system uses copied knowledge embedded in the ML model to replicate expert-level selection accuracy automatically, dramatically reducing time consumption while maintaining precision.

Inventive Principle:
Principle #26Copying

Data Source

PatentEP4332915A1Automated selection and model training for charged particle microscope imaging
Publication Date: 2024.03.06 FEI CO
  • EP4332915A1 patent drawingFigure 1A~1B
  • EP4332915A1 patent drawingFigure 2A
  • EP4332915A1 patent drawingFigure 2B

AI summary

Disclosed herein are CPM support systems, as well as related methods, computing devices, and computer-readable media. For example, in some embodiments, a method may comprise determining, based on selection data indicating selections of areas of microscopy imaging data, training data for a machine-learning model. The method may comprise training, based on the training data, the machine-learning model to automatically determine one or more areas of microscopy imaging data for performing at least one operation, such as high resolution data acquisition and data analysis. The method may comprise causing a computing device to be configured to use the machine-learning model to automatically determine areas of microscopy imaging data for the at least one operation.