BIOFLow International Research Experience for Students (IRES)

Monitoring of PM particles using a smartphone-based DHM and deep learning

Background.

Airborne particulate matter (PM) causes harmful illnesses, such as respiratory, cardiovascular, and carcinoma diseases, becoming a global environmental issue [1]. Recently, air quality monitoring of PM concentration has emerged as an essential preventive measurement technique. However, conventional PM monitoring techniques require specialized equipment and have technical limitations, such as restriction for high concentration conditions and data fluctuations induced by air intake during the sampling period. To overcome these limitations, a hand-held device for monitoring airborne PMs was developed [2, 3]. The smartphone-based digital holographic microscopy (S-DHM) system and deep-learning-based analysis of holographic speckle patterns are combined for handheld PM monitoring. Further investigation is required to selectively measure concentrations of PMs composed of different particle sizes in a highly concentrated and heterogeneously suspended condition.

IRES student involvement.

Mr. Jihwan Kim will work with an IRES student on this project. The IRES student will generate synthetic speckle patterns of PM particles with various sizes, shapes, and concentration distributions. The synthetic speckle patterns and the corresponding physical properties of PMs will be trained by a deep learning network. The results will be validated by applying the trained model to real speckle patterns of PMs.

References

[1] Goldman, G.T., Dominici, F., 2019. Don’t abandon evidence and process on air pollution policy. Science 363, 1398-1400.

[2] Kim, J., Go, T., Lee, S.J., 2021. Accurate real-time monitoring of high particulate matter concentration based on holographic speckles and deep learning. Journal of Hazardous Materials 409, 124637.

[3] Kim, J., Go, T., Lee, S.J., 2021. Volumetric monitoring of airborne particulate matter concentration using smartphone-based digital holographic microscopy and deep learning. Journal of Hazardous Materials 418, 126351.