Droplet Cell Counting via Convolutional Neural Network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for processing droplets in microfluidic systems face challenges in accurately counting cells or other entities within droplets, particularly due to issues like cells becoming stuck to the droplet edge or occluding each other, which affects monoclonality assurance.
Innovation Solution
A method involving capturing a sequence of images of droplets as they pass through a microfluidic channel and processing these images using a convolutional neural network to count the number of cells or entities. This approach exploits the changing orientation and disposition of droplet contents to improve accuracy, allowing for control of microfluidic processes based on the estimated number of cells or entities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single image of a droplet is used for cell counting, then the processing speed is fast, but the accuracy is reduced due to cells becoming stuck to edges or occluding each other
Solution Approach 1:
The system captures multiple images of droplets at different time points before final counting. By preparing and analyzing multiple images in advance as droplets pass through the microfluidic channel, the system accumulates sufficient data to accurately count cells despite transient occlusions or edge-sticking phenomena, thereby improving measurement precision without requiring excessive processing time
2Reliability
If multiple images of droplets are captured and processed to improve cell counting accuracy, then the monoclonality assurance is improved, but the device complexity increases
Solution Approach 1:
The system uses a convolutional neural network that processes multiple captured images and provides feedback on cell presence and orientation. The network analyzes the sequence of images to determine droplet contents, using the changing orientation and disposition of droplet contents over time to make accurate monoclonality determinations, thereby improving reliability through intelligent feedback processing
Solution Approach 2:
Instead of physically manipulating the droplets multiple times, the system creates multiple optical copies (images) of the same droplet at different time points. These digital copies are then processed by the neural network to determine cell counting accuracy, reducing physical device complexity while maintaining high reliability through multiple observational copies
3Productivity
If cells are allowed to remain in droplets for further processing, then the productivity is maintained, but the accuracy of cell counting is reduced due to cell aggregation
Solution Approach 1:
The system performs cell counting analysis while droplets are still in circulation through the microfluidic channel, before they undergo further processing or aggregation. By capturing and analyzing multiple images during this preliminary phase when cells are still dispersed, the system maintains both high productivity and accurate cell counting, avoiding the need to wait for potential aggregation events
Data Source
AI summary
ABSTRACT:An instrument for processing droplets in a microfluidic system. The instrument captures a time sequence of images of a droplet as it passes through a channel in a microfluidic system. The instrument also processes each image of the sequence of images using a convolutional neural network to count a number of cells or other entities visible in each image the droplet. This involves processing the count of the number of cells or other entities visible in each image of the droplet to determine an estimated number of cells or other entities in the droplet. The instrument controls a microfluidic process performed on the droplet, e.g. droplet dispensing, responsive to the estimated number of cells or other entities in the droplet. The instrument uses the changing orientation and disposition of droplet contents in combination with machine learning to improve monoclonality assurance.