An improved U-Net for cell confluence estimation

Abstract

Cell confluence is an important metric to determine the growth and the best harvest time of adherent cells. At present, the evaluation of cell confluence mainly relies on experienced labor, and thus it is not conducive to the automated cell culture. In this paper, we proposed an improved U-Net algorithm (called DU-Net) for the segmentation of adherent cells. First, the general convolution was replaced by the dilated convolution to expand the receptive fields for feature extraction. Then, the convolutional layers were combined with the batch normalization layers to reduce the dependence of the network on initialization. As a result, the segmentation accuracy and Fl-score of the proposed DU-Net for adherent cells with low confluence (<50%) reached 96.94% and 93.87%, respectively, and for those with high confluence (≥50%), they reached 98.63% and 98.98%, respectively. Further, the paired t-test results showed that the proposed DU-Net was statistically superior to the traditional U-Net algorithm.

References

[1] SongN, ScholtemeijerM, ShahK. Mesenchymal stem cell immunomodulation: mechanisms and therapeutic potential[J]. Trends in pharmacological sciences, 2020, 41(9):653-664

[2] BlauH M, DaleyG Q. Stem cells in the treatment of disease[J]. The new England journal of medicine, 2019, 380(18):1748-1760

[3] GhaneN, VardA, TalebiA, et al.. Segmentation of white blood cells from microscopic images using a novel combination of K-Means clustering and modified watershed algorithm[J]. Journal of medical signals & sensors, 2017, 7(2):92-101

[4] ArulmuruganR, AnandakumarH. Region-based seed point cell segmentation and detection for biomedical image analysis[J]. International journal of biomedical engineering and technology, 2018, 27(4):273-289

[5] JoH J, HanJ H, KimY S, et al.. A novel method for effective cell segmentation and tracking in phase contrast microscopic images[J]. Sensor, 2021, 21(10):3516

[6] MotaS M, RogersR E, HaskellA W, et al.. Automated mesenchymal stem cell segmentation and machine learning-based phenotype classification using morphometric and textural analysis[J]. Journal of medical imaging, 2021, 8(1): 014503

[7] ZhaoZ Q, ZhengP, XuS T, et al.. Object detection with deep learning: a review[J]. IEEE transactions on neural networks and learning systems, 2019, 30(11): 3212-3232

[8] TaghanakiS A, AbhishekK, CohenJ P, et al.. Deep semantic segmentation of natural and medical images: a review[J]. Artificial intelligence review, 2021, 54(1):137-178

[9] RawatW, WangZ H. Deep convolutional neural networks for image classification: a comprehensive review[J]. Neural computation, 2017, 29(9):2352-2449

[10] LongJ, ShelhamerE, DarrellT. Fully convolutional networks for semantic segmentation[C], 2015, New York, IEEE: 3431-3440

[11] RonnebergerO, FischerP, BroxT. U-Net: convolutional networks for biomedical image segmentation[C], 2015, Berlin Heidelberg, Springer: 234-241

[12] BadrinarayananV, KendallA, CipollaR. SegNet: a deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE transactions on pattern analysis and machine intelligence, 2017, 39(12):2481-2495

[13] ZhaoH, ShiJ, QiX, et al.. Pyramid scene parsing network[C], 2017, New York, IEEE: 2881-2890

[14] ChenL C, ZhuY K, PapandreouG, et al.. Encoder-decoder with atrous separable convolution for semantic image segmentation[C], 2018, Berlin Heidelberg, Springer: 801-818

[15] FuJ, LiuJ, TianH, et al.. Dual attention network for scene segmentation[C], 2019, New York, IEEE: 3146-3154

[16] SoleimaniS, MirzaeiM, ToncuD C. A new method of SC image processing for confluence estimation[J]. Micron, 2017, 101: 206-212

[17] ShaoY B, XuZ L, YaoT Z. Calculation of cell growth confluence based on region growth[J]. Computer science and application, 2020, 10(2):236-244

[18] WangZ N, ZhanR C, HuY. Automated confluence measurement method for mesenchymal stem cell from brightfield microscopic images[J]. Microscopy and microanalysis, 2021, 27(5):1093-1101

[19] ZhangM, LiX, XuM J, et al.. Automated semantic segmentation of red blood cells for sickle cell disease[J]. IEEE journal of biomedical and health informatics, 2020, 24(11):3095-3102

[20] BiniciR C, AyanzadehA, TöreyinB U, et al.. Automated segmentation of cells in phase contrast optical microscopy time series images[C], 2019, New York, IEEE: 1-4

[21] TsaiH F, GajdaJ, SloanT, et al.. Usiigaci: instance-aware cell tracking in stain-free phase contrast microscopy enabled by machine learning[J]. SoftwareX, 2019, 9: 230-237

[22] AyanzadehA, ÖzuysalY Ö, OkvurD P, et al.. Deep learning based segmentation pipeline for label-free phase-contrast microscopy images[C], 2021, New York, IEEE: 1-4

[23] YU F, KOLTUN V. Multi-scale context aggregation by dilated convolutions[C]//Proceedings of 2016 the Tenth International Conference on Learning Representations, May 2–4, 2016, Caribe Hilton, San Juan, Puerto Rico.

[24] IoffeS, SzegedyC. Batch normalization: accelerating deep network training by reducing internal covariate shift[C], 2015, Pittsburg, IMLS: 448-456

[25] GongW, HanZ B, ZhaoH, et al.. Banking human umbilical cord derived mesenchymal stromal cells for clinical use[J]. Cell transplantation, 2012, 21(1):207-216

[26] CHEN L C, PAPANDREOU G, SCHROFF F, et al. Rethinking atrous convolution for semantic image segmentation: version 3[EB/OL]. (2017-12-05) [2021-08-20]. https://arxiv.org/abs/1706.05587.

[27] LiaoJ C, CaoL B, LiW, et al.. UnetDVH-Linear: linear feature segmentation by dilated convolution with vertical and horizontal kernels[J]. Sensors, 2020, 20(20):5759

[28] SrivastavaN, HintonG, KrizhevskyA, et al.. Dropout: a simple way to prevent neural networks from overfitting[J]. Journal of machine learning research, 2014, 15(1): 1929-1958