Virtual impactor-based label-free particulate matter detection using holography and deep learning
A particulate matter detection device takes holographic images of flowing particulate matter concentrated by a virtual impactor, which selectively slows down and guides larger particles to fly through an imaging window. The flowing particles are illuminated by a pulsed laser diode, casting their inline holograms on a CMOS image sensor in a lens-free mobile imaging device. The illumination contains three short pulses with a negligible shift of the flowing particle within one pulse and triplicate holograms of the same particle are recorded at a single frame revealing different perspectives of each particle. A deep neural network classifies the particles based on the acquired holographic images. The device was tested using different types of pollen and achieved a blind classification accuracy of 92.91%. This mobile and cost-effective device weighs ˜700 g and can be used for label-free sensing and quantification of various bio-aerosols over extended periods.
1 . A particulate matter detection device comprising:
an air sampler having an inlet for receiving air having particles therein and comprising a virtual impactor comprising a main channel intersecting with one or more bypass channels, wherein the main channel comprises an optically transparent imaging window;
a pulsed light source disposed on a first side of the imaging window;
an image sensor disposed on a second, opposing side of the imaging window; and
wherein the image sensor captures holographic image(s) of the particles passing through the imaging window, wherein each holographic image contains one or more holograms cast by the particles at different times in response to one or more pulse(s) of the light source.
2 . The particulate matter detection device of claim 1 , further comprising a computing device comprising image processing software configured to reconstruct the holographic image(s) into one or more focused image(s) and/or defocused image(s).
3 . The particulate matter detection device of claim 2 , further comprising a trained neural network configured to receive the one or more focused image(s) and/or defocused image(s) and classify the particles into one or more classes.
4 . The particulate matter detection device of claim 3 , wherein the particles comprise mold, pollen, or microplastics and the classes corresponding to the particular type of mold, pollen, or microplastics.
5 . A method of using the particulate matter detection device of claim 3 , comprising:
pulling air having particles therein into the inlet of the air sampler;
capturing holographic image(s) of the particles passing through the imaging window; and
classifying the particles passing through the imaging window with the trained neural network.
6 . The method of claim 5 , further comprising displaying the classification result for the particles on a local or remote display associated with the computing device.
7 . The method of claim 5 , wherein the classification comprises a majority voting scheme based on the classification of each of the holograms cast by the particles at different pulses of the light source.
8 . The particulate matter detection device of claim 2 , wherein the computing device further includes a display with a graphical user interface (GUI) configured to control one or more of: air flow rate, fan(s), flow meter(s), the pulsed light source, and the image sensor.
9 . The particulate matter detection device of claim 2 , wherein the image processing software comprises an auto-focusing algorithm and/or a trained neural network that reconstructs the holographic image(s) into one or more focused image(s).
10 . The particulate matter detection device of claim 1 , wherein the main channel is operatively coupled to a first fan and the one or more bypass channels is/are operatively coupled at least one other fan.
11 . The particulate matter detection device of claim 10 , further comprising a first flow meter configured to measure air flow in the main channel and at least one other flow meter configured to measure air flow in the one or more bypass channels.
12 . The particulate matter detection device of claim 11 , further comprising at least one microcontroller operatively connected to the fan(s), the flow meter(s), and the image sensor.
13 . The particulate matter detection device of claim 1 , wherein each holographic image contains a plurality of holograms cast by the particles from a plurality of pulses of the light source.
14 . The particulate matter detection device of claim 1 , wherein the pulsed light source comprises a pulsed laser diode or a pulsed light emitting diode (LED).
15 . The particulate matter detection device of claim 1 , wherein the imaging window has a sensing volume of ≥20 mm 3 .
16 . The particulate matter detection device of claim 1 , wherein the image sensor is located >1 mm from a bottom of the main channel.
17 . A method of classifying particulate matter particles in air comprising:
flowing the air containing particles into an air sampler having an inlet and comprising a virtual impactor comprising a main channel intersecting with one or more bypass channels, wherein the main channel comprises an optically transparent imaging window;
capturing holographic image(s) of the particles passing through the imaging window in the main channel with an image sensor disposed adjacent to the imaging window, wherein each holographic image contains one or more holograms cast by the particles at different times in response to pulses of light from a pulsed light source;
reconstructing the holographic image(s) into corresponding focused image(s) with image processing software; and
inputting the focused image(s) to a trained neural network configured to classify the particulate matter particles passing through the imaging window into one or more classes.
18 . The method of claim 17 , further comprising generating defocused images of the particulate matter particles with the image processing software and wherein the defocused images are input to the trained neural network along with the focused image(s).
19 . The method of claim 17 , wherein the particulate matter particles comprise mold, pollen, or microplastics and the classes corresponding to the particular type of mold, pollen, or microplastics.
20 . The method of claim 17 , wherein the trained neural network is configured to receive the focused image(s) as well as a plurality of defocused images and classify the particulate matter particles into one or more classes.
21 . The method of claim 17 , wherein the image processing software comprises a trained neural network that reconstructs the holographic image(s) into one or more focused image(s) and/or defocused image(s).