Background Subtraction for High-Throughput Sample Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-throughput optical systems for biological and biochemical reactions face challenges in accurately tracking and validating numerous small-volume samples due to manual errors, system noise, and misreading of identifiers, especially with increasing sample volumes and complexity.
Innovation Solution
A computer-implemented method and system for determining the position of a landmark pattern on a substrate, which includes image processing techniques to remove background noise and accurately locate reaction sites, using projections and pedestal removal to enhance image analysis and distinguish between reaction and non-reaction sites.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual tracking and validation of sample identifiers is performed, then operator control and verification are maintained, but labor intensity increases and operator errors occur
Solution Approach 1:
The patent replaces manual mechanical tracking operations with an automated optical imaging system that captures images of sample identifiers and processes them through computer vision algorithms. The system automatically detects, reads, and validates sample identifiers without requiring manual intervention, thereby maintaining reliability while eliminating operator workload and associated errors.
Solution Approach 2:
The system enables self-service by allowing samples to be automatically identified and tracked through their own unique identifiers. The imaging system captures images of the identifiers, and the processing system automatically extracts and validates the identifier information, making the tracking process autonomous without requiring operator involvement.
2Productivity
If automated identifier scanning is implemented, then throughput increases and labor is reduced, but misreading or unreadable identifiers occur
Solution Approach 1:
The system performs preliminary imaging of sample identifiers before processing. By capturing high-quality images of the identifiers in advance and processing them through robust computer vision algorithms, the system ensures accurate reading even with challenging identifier conditions, thereby maintaining precision while enabling high throughput.
Solution Approach 2:
The patent introduces an intermediary image capture step between the identifier and the reading process. Instead of direct optical reading that may fail with unreadable identifiers, the system captures images as an intermediary representation, which can then be processed with error correction and enhancement algorithms to ensure accurate identification.
3Measurement precision
If background noise is not removed from images, then image processing is simpler and faster, but accurate analysis of small-volume samples becomes impossible
Solution Approach 1:
The patent applies segmentation by dividing the image processing into distinct stages: background noise removal, signal enhancement, and feature extraction. By segmenting the processing pipeline, the system can effectively remove background noise to improve measurement precision while managing complexity through modular, organized processing steps.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for improving image quality is provided. The method includes receiving image data of a substrate, wherein the image data is generated by imaging the substrate, and an image is generated from the image data. The method further includes generating a background representation from a background noise portion of the image, wherein the background portion includes signal information undesired for further processing and generating a background subtracted image by subtracting the background representation from the image. In this way, a separate background image is not needed to subtract the background from the image including the regions-of-interest to improve image quality.