Charged Particle Microscope Image Quality Estimation via Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current sample observation devices face challenges in achieving high-quality images with higher throughput due to issues like degraded resolution, signal-to-noise ratio, blurring, and image shaking, especially under higher throughput imaging conditions, where focus misalignment and image distortion occur, and existing methods struggle to accurately estimate high-quality images from blurred images.
Innovation Solution
A sample observation device and method that utilize a charged particle microscope, image storage, and arithmetic units to calculate and apply estimation process parameters for enhancing image quality from degraded images, including learning techniques to estimate high-quality images from paired low and high-quality images, and using convolution neural networks to process images and correct for focus misalignment and distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the number of added frames is reduced to improve throughput, then productivity increases, but image quality (signal-to-noise ratio) deteriorates
Solution Approach 1:
The system performs preliminary actions by acquiring multiple frames at reduced dose before final image reconstruction, preparing data in advance that will be processed to achieve both high throughput and high image quality through the learning model
Solution Approach 2:
A learning model acts as an intermediary between the degraded low-dose images and the desired high-quality images, transferring knowledge from high-dose training images to enable quality enhancement without requiring high dose during actual operation
2Measurement precision
If scanning speed is reduced to improve signal-to-noise ratio, then image quality improves, but image acquisition time increases
Solution Approach 1:
The system changes the dose parameter by acquiring images at reduced dose (higher scanning speed) while using a learning model to compensate for the quality loss, thereby maintaining image quality without sacrificing acquisition speed
3Productivity
If focus height is not adjusted to coincide with sample surface, then throughput increases, but image quality (blurring) deteriorates
Solution Approach 1:
The system converts the harmful effect of focus misalignment and motion blur into a beneficial training opportunity by using these degraded images as input for the learning model, which learns to reconstruct sharp images from blurred inputs, thereby enabling high throughput operation without quality loss
4Productivity
If stage movement speed is increased to improve throughput, then productivity increases, but image stability (shaking) deteriorates
Solution Approach 1:
The learning model serves as an intermediary that stabilizes the imaging system by learning the relationship between motion-blurred images (from fast stage movement) and sharp reference images, enabling the system to compensate for motion-induced instability and maintain image quality at high throughput
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables the accurate estimation and output of high-quality images from degraded images, improving throughput and image quality while reducing measurement errors across different imaging devices, by learning and estimating image quality differences between devices.
Implementation Method 1
a charged particle microscope that images a sample placed on a movable table by irradiating and scanning the sample with a charged particle beam
Data Source
AI summary
A sample observation device images a sample placed on a movable table by irradiating and scanning the sample with a charged particle beam of a microscope. A degraded image having poor image quality and a high quality image having satisfactory image quality which are acquired at the same location of the sample by causing the charged particle microscope to change an imaging condition for imaging the sample are stored. An estimation process parameter is calculated for estimating the high quality image from the degraded image by using the stored degraded image and high quality image. A high quality image estimation unit processes the degraded image obtained by causing the charged particle microscope to image the desired site of the sample by using the calculated estimation process parameter. Thereby, the high quality image obtained at the desired site is estimated, and then the estimated high quality image is output.