Classifying microfluidic droplets is an essential step in imaging flow cytometry. While deep learning algorithms can detect and classify such droplets in benchtop laboratory settings, their deployment on portable devices remains challenging because the computational requirements often exceed the capabilities of compact, resource-limited devices. This hinders the transition from stationary lab setups to field-deployable instruments. To tackle this issue, we introduce a customized YoloV4-tiny model deployed on a Raspberry Pi-5 (RPi5) single-board computer. Our neural network is trained using 878 images from a custom dataset of 975 images, derived from two videos captured with real-life microfluidic experimental setup. We evaluate performance based on inference time and mean average precision. Our system successfully classifies three distinct droplet types (no cell, one cell, multiple cells) within 13 ms, achieving a 99.95% mean average precision at an intersection over union threshold of 0.5 (mAP@0.5). We also compare the classification performance metrics of our customized YoloV4-tiny model against seven other combinations of machine learning models and platforms, including a recent low-cost, highly compact edge device with tensor processing unit capabilities, specifically, the MaixCam board with LicheeRV Nano module (SOPHGO SG2002) running a YoloV5-s model. Compared to this proposed customized YoloV4-tiny on the RPi5, the YoloV5-s on MaixCam achieves a significantly shorter classification time (5.34 ms) owing to its onboard tensor processing unit but suffers from a lower mAP@0.5 of 55.09% due to quantization. Our work shows that carefully designed systems can achieve a balance between speed and accuracy, enabling robust performance even on resource-limited devices and paving the way for microfluidic droplet classification in portable imaging flow cytometry.
1. Trinh, T. N. D., Do, H. D. K., Nam, N. N., Dan, T. T., Trinh, K. T. L. and Lee, N. Y. Droplet-based microfluidics: applications in pharmaceuticals. Pharmaceuticals, 2023, 16(7), 937.
https://doi.org/10.3390/ph16070937
2. Chen, Z., Kheiri, S., Young, E. W. K. and Kumacheva, E. Trends in droplet microfluidics: from droplet generation to biomedical applications. Langmuir, 2022, 38(20), 6233–6248.
https://doi.org/10.1021/acs.langmuir.2c00491
3. Liu, Z., Jin, L., Chen, J., Fang, Q., Ablameyko, S., Yin, Z. et al. A survey on applications of deep learning in microscopy image analysis. Comput. Biol. Med., 2021, 134, 104523.
https://doi.org/10.1016/j.compbiomed.2021.104523
4. Rutkowski, G. P., Azizov, I., Unmann, E., Dudek, M. and Grimes, B. A. Microfluidic droplet detection via region-based and single-pass convolutional neural networks with comparison to conventional image analysis methodologies. Mach. Learn. Appl., 2022, 7, 100222.
https://doi.org/10.1016/j.mlwa.2021.100222
5. Afrin, F., Le Moullec, Y. and Pardy, T. Microfluidic droplet classification through tuned convolutional neural network on a resource constrained platform. In 2024 19th Biennial Baltic Electronics Conference (BEC), Tallinn, Estonia, 2–4 October 2024. IEEE, 2024, 1–4.
https://doi.org/10.1109/BEC61458.2024.10737958
6. Zhou, X., Mao, Y., Gu, M. and Cheng, Z. WSCNet: biomedical image recognition for cell encapsulated microfluidic droplets. Biosensors, 2023, 13(8), 821.
https://doi.org/10.3390/bios13080821
7. Soldati, G., Del Ben, F., Brisotto, G., Biscontin, E., Bulfoni, M., Piruska, A. et al. Microfluidic droplets content classification and analysis through convolutional neural networks in a liquid biopsy workflow. Am. J. Transl. Res., 2018, 10(12), 4004–4016.
8. Kensert, A., Harrison, P. J. and Spjuth, O. Transfer learning with deep convolutional neural networks for classifying cellular morphological changes. SLAS Discov., 2019, 24(4), 466–475.
https://doi.org/10.1177/2472555218818756
9. Lee, K., Kim, S.-E., Doh, J., Kim, K. and Chung, W. K. User-friendly image-activated microfluidic cell sorting technique using an optimized, fast deep learning algorithm. Lab Chip, 2021, 21(9), 1798–1810.
https://doi.org/10.1039/D0LC00747A
10. Xu, J., Fan, W., Madsen, J., Tanev, G. P. and Pezzarossa, L. AI-based detection of droplets and bubbles in digital microfluidic biochips. In 2023 Design, Automation & Test in Europe Conference & Exhibition (DATE), Antwerp, Belgium, 17–19 April 2023. IEEE, 2023, 1–6.
https://doi.org/10.23919/DATE56975.2023.10136887
11. Ghafari, M., Mailman, D., Hatami, P., Peyton, T., Yang, L. and Dang, W. A comparison of YOLO and Mask-RCNN for detecting cells from microfluidic images. In 2022 International Conference on Artificial Intelligence in Information and Communication (ICAIIC), Jeju Island, Korea, 21–24 February 2022. IEEE, 2022, 204–209.
https://doi.org/10.1109/ICAIIC54071.2022.9722616
12. Li, Y., Mahjoubfar, A., Chen, C. L., Niazi, K. R., Pei, L. and Jalali, B. Deep cytometry: deep learning with real-time inference in cell sorting and flow cytometry. Sci. Rep., 2019, 9, 11088.
https://doi.org/10.1038/s41598-019-47193-6
13. Suzuki, Y., Kobayashi, K., Wakisaka, Y. and Ozeki, Y. Label-free chemical imaging flow cytometry by high-speed multicolor stimulated Raman scattering. Proc. Natl. Acad. Sci. U.S.A., 2019, 116(32), 15842–15848.
https://doi.org/10.1073/pnas.1902322116
14. Meng, N., Lam, E. Y., Tsia, K. K. and So, H. K.-H. Large-scale multi-class image-based cell classification with deep learning. IEEE J. Biomed. Health Inform., 2019, 23(5), 2091–2098.
https://doi.org/10.1109/jbhi.2018.2878878
15. Riti, J., Sutra, G., Naas, T., Volland, H., Simon, S. and Perez-Toralla, K. Combining deep learning and droplet microfluidics for rapid and label-free antimicrobial susceptibility testing of colistin. Biosens. Bioelectron., 2024, 257, 116301.
https://doi.org/10.1016/j.bios.2024.116301
16. Ahmadpour, A., Shojaeian, M. and Tasoglu, S. Deep learning-augmented T-junction droplet generation. iScience, 2024, 27(4), 109326. https://doi.org/10.1016/j.isci.2024.109326
17. Asama, R., Liu, C. J. S., Tominaga, M., Cheng, Y.-R., Nakamura, Y., Kondo, A. et al. Droplet-based microfluidic platform for detecting agonistic peptides that are self-secreted by yeast expressing a G-protein-coupled receptor. Microb. Cell Factories, 2024, 23, 104.
http://dx.doi.org/10.1186/s12934-024-02379-0
18. Raspberry. Raspberry Pi Hardware.
https://www.raspberrypi.com/documentation/computers/raspberry-pi.html (accessed 2024-12-13).
19. Khan, S. Z., Le Moullec, Y. and Alam, M. M. An NB-IoT-based edge-of-things framework for energy-efficient image transfer. Sensors, 2021, 21(17), 5929.
https://doi.org/10.3390/s21175929
20. Sipeed. MaixCam – Fast Development for AI Vision and Audio Projects.
https://wiki.sipeed.com/hardware/en/maixcam/index.html (accessed 2024-12-13).
21. De Jonghe, J., Kaminski, T. S., Morse, D. B., Tabaka, M., Ellermann, A. L., Kohler, T. N. et al. spinDrop: a droplet microfluidic platform to maximise single-cell sequencing information content. Nat. Commun., 2023, 14, 4788.
http://dx.doi.org/10.1101/2023.01.12.523500
22. Skalski, P. makesense.ai.
https://github.com/SkalskiP/make-sense (accessed 2024-12-14).
23. Redmon, J. Darknet Open Source Neural Network Framework.
https://github.com/pjreddie/darknet (accessed 2024-12-14).
24. Sipeed. LicheeRV Nano.
https://wiki.sipeed.com/hardware/en/lichee/RV_Nano/1_intro.html (accessed 2024-12-13).
25. SOPHGO. SG200X Hardware.
https://github.com/sophgo/sophgo-hardware/tree/master/SG200X (accessed 2024-12-13).
26. Sipeed. MaixHub.
https://maixhub.com/ (accessed 2024-12-14).
27. Jõemaa, R., Gyimah, N., Ashraf, K., Pärnamets, K., Zaft, A. and Scheler, O. CogniFlow-Drop: integrated modular system for automated generation of droplets in microfluidic applications. IEEE Access, 2023, 11, 104905–104929.
https://doi.org/10.1109/ACCESS.2023.3316726
