Charged Particle Microscopy Masking for Selective Data Acquisition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional charged particle microscopy (CPM) techniques require significant user supervision and a heavy upfront investment of time and energy for training, leading to inefficiencies in imaging throughput, radiation damage, and data storage requirements.
Innovation Solution
A CPM data acquisition system utilizing machine-learning models trained with minimal user input to generate selective imaging masks, reducing the need for extensive training datasets and enabling faster, more efficient data acquisition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional CPM techniques are used with extensive training datasets and user supervision, then measurement precision and reliability are improved, but productivity and time efficiency deteriorate
Solution Approach 1:
The system performs preliminary action by pre-training machine learning models on large datasets before actual microscopy operation. This preliminary training enables the model to automatically recognize features during imaging without requiring real-time user supervision or extensive per-sample training, thereby resolving the contradiction between measurement precision and productivity
Solution Approach 2:
The machine learning model performs self-service by automatically identifying features of interest and generating masks without continuous user intervention. The model serves itself by making autonomous decisions about which regions to image at high resolution, eliminating the need for user supervision while maintaining accurate feature recognition
2Measurement precision
If conventional CPM techniques acquire full-field high-resolution images, then measurement precision is improved, but radiation damage and data storage requirements worsen
Solution Approach 1:
The system applies local quality by using machine learning models to identify specific regions of interest and generate masks that indicate only those areas require high-resolution imaging. This allows the charged particle beam to focus radiation only on relevant local areas rather than illuminating the entire field, thereby maintaining measurement precision while significantly reducing radiation damage to the specimen
Solution Approach 2:
The system uses partial action by acquiring high-resolution images only of selected regions identified by the machine learning model, rather than imaging the entire field at high resolution. The mask generated by the model guides the beam to apply partial illumination only where needed, reducing overall radiation exposure while preserving critical measurement precision for features of interest
3Productivity
If machine learning models are trained with minimal user input, then productivity and ease of operation are improved, but measurement precision may deteriorate
Solution Approach 1:
The system performs preliminary action by pre-training machine learning models on large, diverse datasets before deployment. This extensive preliminary training compensates for minimal user input during actual operation, ensuring the model achieves high measurement precision while maintaining fast productivity during imaging through automatic feature recognition without requiring users to curate training data
Data Source
AI summary
Disclosed herein are charged particle microscopy (CPM) support systems, as well as related methods, computing devices, and computer-readable media. For example, in some embodiments, a CPM support apparatus may include: first logic to cause a CPM to generate a single image of a first portion of a specimen; second logic to generate a first mask based on one or more regions-of-interest provided by user annotation of the single image; and third logic to train a machine-learning model using the single image and the one or more regions-of-interest. The first logic may cause the CPM to generate multiple images of corresponding multiple additional portions of the specimen, and the second logic may, after the machine-learning model is trained using the single image and the one or more regions-of-interest, generate multiple masks based on the corresponding images of the additional portions of the specimen using the machine-learning model without retraining.