Portable Imaging Flow Cytometer With Deep Learning Phase Recovery

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing imaging flow cytometers face limitations in throughput, image quality, and cost due to the use of microscope objectives, which restrict field-of-view and depth-of-field, making them expensive and less suitable for high-resolution, high-throughput analysis of plankton and microorganisms in water samples.

Innovation Solution

A compact, cost-effective, and portable in-line holographic imaging flow cytometer using deep learning-enabled phase recovery and holographic reconstruction, capable of capturing high-resolution, color images of label-free objects in real-time at a throughput of ~100 mL/h, without fluorescence triggering or hydrodynamic focusing, utilizing a microfluidic chip and a computing device for image processing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a microscope objective lens is used to image plankton in flow, then image resolution is improved, but volumetric throughput is limited to a few mL per hour

Engineering Contradiction:
Improveimage resolutionVSAvoidvolumetric throughput
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent removes the microscope objective lens from the optical path, extracting the imaging function from traditional flow cytometry. This allows direct imaging of the flow channel without the resolution-throughput trade-off imposed by objectives, enabling both high resolution and high throughput simultaneously

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent transitions from point-by-point scanning imaging (traditional microscopy) to planar parallel imaging by capturing the entire flow channel cross-section in a single camera frame. This dimensional change from 1D scanning to 2D parallel acquisition enables throughput increase while maintaining resolution

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Measurement precision

If a microscope objective lens is used, then image quality is improved, but device cost increases

Engineering Contradiction:
Improveimage qualityVSAvoiddevice cost
Core Design Contradiction:
Measurement precisionVSEase of manufacture

Solution Approach 1:

The patent replaces expensive, precision-engineered microscope objective lenses with inexpensive, off-the-shelf components including a simple camera and transparent plate. This substitution dramatically reduces device cost while maintaining adequate image quality for plankton identification

Inventive Principle:
Principle #27Cheap short-living objects (Disposable)

Solution Approach 2:

The patent creates a direct optical copy of the flow channel contents onto the camera sensor without requiring complex objective lenses. The transparent plate serves as a simple interface that copies the microscopic scene directly to the sensor plane, eliminating the need for expensive optical components

Inventive Principle:
Principle #26Copying

3Measurement precision

If hydrodynamic focusing is used to confine sample to focal point, then measurement precision is improved, but device complexity increases

Engineering Contradiction:
Improvepositioning precisionVSAvoidfluidic system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent eliminates the need for complex hydrodynamic focusing systems by allowing plankton to pass through the imaging region in natural flow. The transparent plate creates a defined imaging plane that works with natural flow patterns, requiring no additional focusing channels or pressure control mechanisms

Inventive Principle:
Principle #25Self-service

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

The device achieves high-throughput imaging of microorganisms, including plankton and parasites, with improved image quality and cost-effectiveness, enabling continuous monitoring of water bodies, and is validated by field tests showing good agreement with independent measurements.

Implementation Method 1

A light source (e.g., LED) is disposed within the housing or enclosure and used to provide illumination

Methodology Applied
Scientific EffectLight: Light

Implementation Method 2

an image sensor disposed adjacent to the microfluidic channel and within an optical path that receives light from the light source that passes through the microfluidic channel, the image sensor configured to capture a plurality of image frames containing raw hologram images of the objects

Methodology Applied
Scientific EffectHolography: Interference

Data Source

PatentEP3803384B1Deep learning-enabled portable imaging flow cytometer for label-free analysis of water samples
Publication Date: 2025.08.06 RGT UNIV OF CALIFORNIA
  • EP3803384B1 patent drawingFigure 1~2
  • EP3803384B1 patent drawingFigure 3
  • EP3803384B1 patent drawingFigure 4A

AI summary

An imaging flow cytometer device includes a housing holding a multi-color illumination source configured for pulsed or continuous wave operation. A microfluidic channel is disposed in the housing and is fluidically coupled to a source of fluid containing objects that flow through the microfluidic channel. A color image sensor is disposed adjacent to the microfluidic channel and receives light from the illumination source that passes through the microfluidic channel. The image sensor captures image frames containing raw hologram images of the moving objects passing through the microfluidic channel. The image frames are subject to image processing to reconstruct phase and/or intensity images of the moving objects for each color. The reconstructed phase and/or intensity images are then input to a trained deep neural network that outputs a phase recovered image of the moving objects. The trained deep neural network may also be trained to classify object types.