Performance prediction of variable-width microfluidic concentration gradient chips by deep learning

doi:10.16085/j.issn.1000-6613.2022-1839

Abstract

Abstract:

With the development of personalized medicine, customized medications are receiving increasing attention. In order to produce customized medications, it is necessary to prepare medication mixture solutions of specified concentrations. We proposed the design of random variable-width (RVW) microfluidic chips, and predicted their performance through Convolutional Neural Networks. First, a design scheme of RVW microchannel was proposed, and the outlet concentrations and the outlet flow rates were obtained by simulation. Second, the KD-MiniVGGNet model was designed according to the principle of convolutional kernel decomposition. The model was trained with the concentration and flow rate data and predicted the outlet concentration and outlet flow rate for more concentration gradient chips. Finally, an experimental research system was built to verify the accuracy of the prediction results of the KD-MiniVGGNet model. The results showed that the RVW microfluidic concentration gradient chips could widen the range of outlet flow rates by 66.7%. When the query conditions were the same, the RVW concentration gradient chip widened the distribution range of outlet concentration of the three outlets by 9%, 16% and 11%, and the distribution range of outlet velocity of the three outlets by 29%, 28% and 30%, respectively. The accuracy of KD-MiniVGGNet model on the test set of outlet concentrations and flow rates could reach 91.5% and 92.7%, respectively. The average absolute error between the prediction results of KD-MiniVGGNet model and the experimental results was 4.3%. The design method proposed in this study could achieve efficient and accurate design of concentration gradient chips, optimize the performance range of concentration gradient chips, and better offer solution preparation services for pharmaceutical customization.

Key words: microchannels, random variable-width microchannels, numerical simulation, neural network, medicine solution preparation

摘要：

随着个性化医疗的发展，定制化药物受到了越来越多的关注，为了生产定制化药物，需要制备指定浓度的药物混合溶液。本研究首次提出了一种随机变宽度（RVW）结构的微流控浓度梯度芯片，并通过卷积神经网络算法实现芯片的性能预测。首先，设计了一种RVW微流道结构并通过仿真模拟得到出口浓度和出口流速。其次，根据卷积核分解原理设计了KD-MiniVGGNet深度学习模型，使用仿真模拟得到的浓度和流速数据训练模型并预测更多浓度梯度芯片的出口浓度和出口流速。最后，通过实验验证了KD-MiniVGGNet深度学习模型预测结果的准确性。研究结果表明：相较于随机等宽度（REW）浓度梯度芯片，RVW浓度梯度芯片的出口集中流速范围提高了66.7%，三个出口的出口浓度分布范围分别拓宽了9%、16%和11%，三个出口的出口流速分布范围分别拓宽了29%、28%和30%；KD-MiniVGGNet模型在出口浓度和出口流速测试集上的模型准确率分别达到91.5%和92.7%；出口浓度的KD-MiniVGGNet模型预测结果与实验结果之间的平均误差为4.3%。本研究中所提出的设计方法可提高浓度梯度芯片结构的多样性，进一步优化浓度梯度芯片的性能范围，更好地为药物定制提供溶液制备服务。

关键词: 微通道, 随机变宽度微流道, 数值模拟, 神经网络, 药物溶液制备

CLC Number:

TQ050.2

YU Junnan, YU Jianfeng, CHENG Yang, QI Yibo, HUA Chunjian, JIANG Yi. Performance prediction of variable-width microfluidic concentration gradient chips by deep learning[J]. Chemical Industry and Engineering Progress, 2023, 42(7): 3383-3393.

俞俊楠, 俞建峰, 程洋, 齐一搏, 化春键, 蒋毅. 基于深度学习的变宽度浓度梯度芯片性能预测[J]. 化工进展, 2023, 42(7): 3383-3393.

Figures/Tables 17

References 28

1	GOETZ L H, SCHORK N J. Personalized medicine: Motivation, challenges, and progress[J]. Fertility and Sterility, 2018, 109(6): 952-963.
2	VAZ V M, KUMAR L. 3D printing as a promising tool in personalized medicine[J]. AAPS PharmSciTech, 2021, 22(1): 1-20.
3	WYATT SHIELDS IV C, REYES C D, LÓPEZ G P. Microfluidic cell sorting: A review of the advances in the separation of cells from debulking to rare cell isolation[J]. Lab on a Chip, 2015, 15(5): 1230-1249.
4	CHI Chunwei, AHMED A H R, DERELI-KORKUT Z, et al. Microfluidic cell chips for high-throughput drug screening[J]. Bioanalysis, 2016, 8(9): 921-937.
5	ZHANG Shuyuan, LIANG Xiao, HUANG Xinye, et al. Precise and fast microdroplet size distribution measurement using deep learning[J]. Chemical Engineering Science, 2022, 247: 116926.
6	GOŁĄB M, WOZNIAKIEWICZ M, NOWAK P M, et al. An automated hydrodynamically mediated technique for preparation of calibration solutions via capillary electrophoresis system as a promising alternative to manual pipetting[J]. Molecules (Basel, Switzerland), 2021, 26(20): 6268.
7	FLEISCHER H, BAUMANN D, JOSHI S, et al. Analytical measurements and efficient process generation using a dual-arm robot equipped with electronic pipettes[J]. Energies, 2018, 11(10): 2567.
8	CAI Gaozhe, XUE Li, ZHANG Huilin, et al. A review on micromixers[J]. Micromachines, 2017, 8(9): 274.
9	王炳捷, 李辉, 杨晓勇, 等. CFD数值模拟技术在液滴微流控多相流特性研究的应用进展[J]. 化工进展, 2021, 40(4): 1715-1735.
	WANG Bingjie, LI Hui, YANG Xiaoyong, et al. Application process of CFD-numerical simulation technology for multiphase flow characteristics study in droplet-microfluidic systems[J]. Chemical Industry and Engineering Progress, 2021, 40(4): 1715-1735.
10	CHANG Bo, KIVINEN O, PINI I, et al. Nanoliter deposition on star-shaped hydrophilic-superhydrophobic patterned surfaces[J]. Soft Matter, 2018, 14(36): 7500-7506.
11	ELVIRA K S, XAVIER CASADEVALL I S, WOOTTON R C R, et al. The past, present and potential for microfluidic reactor technology in chemical synthesis[J]. Nature Chemistry, 2013, 5(11): 905-915.
12	BEIN A, SHIN Woojung, JALILI-FIROOZINEZHAD S, et al. Microfluidic organ-on-a-chip models of human intestine[J]. Cellular and Molecular Gastroenterology and Hepatology, 2018, 5(4): 659-668.
13	LEE Sanghyun, KIM Hojin, LEE Wonhyung, et al. Microfluidic-based cell handling devices for biochemical applications[J]. Journal of Micromechanics and Microengineering, 2018, 28(12): 123001.
14	GENG Chunyang, LI Chiyu, LI Wang, et al. A simple fabricated microfluidic chip for urine sample-based bladder cancer detection[J]. Journal of Micromechanics and Microengineering, 2018, 28(11): 115011.
15	LAXMI V, JOSHI S S, AGRAWAL A. Extracting white blood cells from blood on microfluidics platform: A review of isolation techniques and working mechanisms[J]. Journal of Micromechanics and Microengineering, 2022, 32(5): 053001.
16	WALKER G M, OZERS M S, BEEBE D J. Cell infection within a microfluidic device using virus gradients[J]. Sensors and Actuators B: Chemical, 2004, 98(2/3): 347-355.
17	SAAD M G, SELAHI A, ZOROMBA M S, et al. A droplet-based gradient microfluidic to monitor and evaluate the growth of Chlorella vulgaris under different levels of nitrogen and temperatures[J]. Algal Research, 2019, 44: 101657.
18	关尧. 面向药物微球的浓度微液滴芯片的操控研究[D]. 杭州: 杭州电子科技大学, 2021.
	GUAN Yao. Research on manipulation of concentration micro-droplet microfluidic chips for drug microspheres[D]. Hangzhou: Hangzhou Dianzi University, 2021.
19	LASHKARIPOUR A, RODRIGUEZ C, MEHDIPOUR N, et al. Machine learning enables design automation of microfluidic flow-focusing droplet generation[J]. Nature Communications, 2021, 12(1): 1-14.
20	LASHKARIPOUR A, GOHARIMANESH M, ABOUEI MEHRIZI A, et al. An adaptive neural-fuzzy approach for microfluidic droplet size prediction[J]. Microelectronics Journal, 2018, 78: 73-80.
21	杨文宇, 谢应明, 闫坤, 等. 基于灰色关联BP神经网络的压缩式蓄冷系统中的水合物生成量预测[J]. 化工进展, 2021, 40(2): 664-670.
	YANG Wenyu, XIE Yingming, YAN Kun, et al. Prediction of hydrate production in compressive cold storage system based on grey relational BP neural network[J]. Chemical Industry and Engineering Progress, 2021, 40(2): 664-670.
22	RAWAT W, WANG Zenghui. Deep convolutional neural networks for image classification: A comprehensive review[J]. Neural Computation, 2017, 29(9): 2352-2449.
23	KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60(6): 84-90.
24	WANG Junchao, BRISK P, GROVER W H. Random design of microfluidics[J]. Lab on a Chip, 2016, 16(21): 4212-4219.
25	JI Weiqing, Tsung Yi HO, WANG Junchao, et al. Microfluidic design for concentration gradient generation using artificial neural network[J]. IEEE Transactions on Computer: Aided Design of Integrated Circuits and Systems, 2020, 39(10): 2544-2557.
26	WANG Junchao, ZHANG Naiyin, CHEN Jinkai, et al. Predicting the fluid behavior of random microfluidic mixers using convolutional neural networks[J]. Lab on a Chip, 2021, 21(2): 296-309.
27	ROSEBROCK A. Deep learning for computer vision with python[M]. Midtown Manhattan: W. W. Norton & Company, 2017: 229-238.
28	SZEGEDY C, VANHOUCKE V, IOFFE S, et al. Rethinking the inception architecture for computer vision[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2016: 2818-2826.

模拟参数	数值
流体密度ρ/kg·m^-3	10³
流体动力黏度μ/Pa·s^-1	10^-3
分子扩散系数D/m²·s^-1	10^-9
出口压力p/Pa	0
壁面流速u/m·s^-1	0

模拟参数	数值
流体密度ρ/kg·m^-3	10³
流体动力黏度μ/Pa·s^-1	10^-3
分子扩散系数D/m²·s^-1	10^-9
出口压力p/Pa	0
壁面流速u/m·s^-1	0

阶段	类型	卷积核尺寸	输出通道数
阶段一	Conv1	3×3	64
	Conv2	3×3	64
阶段二	Conv3	3×3	128
	Conv4	3×3	128
阶段三	FC1	—	128
	FC2	—	3或2^①

阶段	类型	卷积核尺寸	输出通道数
阶段一	Conv1	3×3	64
	Conv2	3×3	64
阶段二	Conv3	3×3	128
	Conv4	3×3	128
阶段三	FC1	—	128
	FC2	—	3或2^①

阶段	类型	卷积核尺寸	输出通道数
阶段一	Conv1	3×3	32
	Conv2~Conv9	2×2	32
阶段二	Conv10	3×3	64
	Conv11~Conv12	2×2	64
阶段三	FC1	—	64
	FC2	—	3或2^①