AI Super-Resolution for Wide-Field 3D Refractive Index Microscopy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3-D refractive index microscopes face a trade-off between wide field of view and high resolution, as increasing numerical aperture reduces the measurable field of view, and existing methods to enhance resolution are time and calculation intensive.
Innovation Solution
A system and method using artificial intelligence (AI) technology, specifically deep learning algorithms, to convert low-resolution 3-D refractive index images to high-resolution images by extracting and enhancing 3-D refractive index patches, utilizing a convolutional neural network (CNN) for efficient resolution improvement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a lens with high numerical aperture (NA) is used to achieve high resolution, then space resolution is improved, but the field of view (FoV) is reduced
Solution Approach 1:
The patent divides the image processing into patches and processes them separately through the neural network. The low-resolution image is segmented into multiple patches, each of which is then super-resolved independently, allowing the system to handle wide-field images while maintaining high resolution through localized processing
Solution Approach 2:
The patent replaces the traditional optical-mechanical approach (using high-NA lenses with limited FoV) with an AI-based computational approach. Instead of relying on physical optical components to achieve high resolution, a deep learning model is trained to predict high-resolution images from low-resolution inputs, eliminating the need for complex optical systems with restricted fields of view
2Area of stationary object
If a lens with low numerical aperture (NA) is used to achieve wide field of view, then the measurable area is improved, but space resolution is reduced
Solution Approach 1:
The patent changes the resolution parameter of the input image through AI processing. The low-resolution image captured by the low-NA lens is transformed into a high-resolution image by the neural network, which learns the mapping between low and high resolution representations during training, thereby improving spatial resolution without sacrificing field of view
Solution Approach 2:
The patent substitutes the mechanical/optical resolution enhancement methods with an AI-based computational method. Instead of using complex optical systems or physical scanning mechanisms, a deep learning model is employed to predict and generate high-resolution images from low-resolution inputs, achieving resolution improvement without mechanical complexity
3Measurement precision
If existing methods are used to increase resolution by measuring images with several incident angles or scanning, then resolution is improved, but time consumption and calculation complexity increase
Solution Approach 1:
The patent performs preliminary training of the neural network model using paired low-resolution and high-resolution images. During this offline training phase, the model learns the complex mapping relationships between different resolution levels. Once trained, the model can rapidly predict high-resolution images from low-resolution inputs without requiring time-consuming multi-angle measurements or scanning operations during actual use
Solution Approach 2:
The patent replaces time-consuming physical measurement methods (multi-angle imaging, scanning) with a computational AI approach. The deep learning model, trained offline, performs resolution enhancement through rapid image processing algorithms, eliminating the need for lengthy physical measurement procedures and reducing both time consumption and calculation complexity during operation
Data Source
AI summary
The present disclosure provides a system and method for improving image resolution of a three-dimensional (3-D) refractive index microscope based on an artificial intelligence (AI) technology. The present disclosure provides a technology for converting a low-resolution 3-D refractive index microscope image into a high-resolution 3-D refractive index image without physical machine conversion and re-photographing based on AI. That is, the present disclosure applies an AI technology, such as deep learning, in order to train an AI model with a physical correlation between a low-resolution 3-D refractive index microscope image and a high-resolution 3-D refractive index image of various samples, such as a cell and a tissue, and convert a low resolution image into a high resolution image without a change in a physical microscope based on the training. Furthermore, for the training of the AI model, the present disclosure uses physical characteristics of a refractive index image.