4. Discussion
Yeasts are unicellular organisms that are widely used in genetics, cell biology and industrial fermentation. The automatic counting method used in this research can be applied to multiple kinds of yeasts, such asSaccharomyces cerevisiae , Schizosaccharomyces pombe , andPichia pastoris . Compared with manual counting and expensive cell counting instruments, the automated method described here is convenient, rapid, reproducible and extremely low cost. An ordinary microscope that can take photographs, as well as the freeware ilastik and ImageJ, are the only tools used in the method, which is easy to manipulate after a short training time, saving both time and reducing human subjective evaluation. In addition, the method can also be widely used in other fungal experiments, such as counting fungal spores when calculating the spore germination rate of filamentous fungi. Organelles and cell structures can also be quantitatively analysed when combined with the use of fluorescent dyes and fluorescent fusion proteins. As a consequence, this method is simple and flexible and has a wide range of applications.
The combination of ilastik and ImageJ in yeast cell counting gives full play to their respective advantages, in which ilastik is used for segmentation and classification based on deep learning and an ImageJ macro performs cell counting by adjusting various parameters. ImageJ also has the ability to distinguish between background and cells (Plugins→Segmentation→Trainable Weka Segmentation). However, in the processing process of ImageJ, the efficiency of dealing with a single image is much lower than that of ilastik. Moreover, Fiji cannot learn multiple images at one time. Thus, ilastik is the optimal choice for segmentation with deep learning-based high accuracy and high throughput. For the freeware ilastik, annotations of background (Lable1) and individual objects (Lable2) serve as inputs into a regression random forest that estimates the object density in every pixel of the image [20]. However, the ”Cell Density Counting” function in ilastik cannot be directly used to obtain an accurate number of yeast cells. Taking the capsule-surrounded C. deneoformans as an example, multiple bright spots in the background and the halo of light generated by the capsule around the yeast cell would be recognized as the same pixel as the cells and thus entered into the analysis of cell numbers. This makes the number of yeast cells counted by ilastik much higher than the actual number.
However, there is no doubt that this method has some problems; for example, different batches of image data may have different background depths and complexities. At the time, it is necessary for ilastik software to distinguish and learn each of the different batches of images as it cannot automatically homogenize the pictures, which can be used as an improvement point in future software development. However, what is certain is that the combination of ImageJ and ilastik can greatly reduce the experimental time and lead to greater accuracy when a large number of yeast samples need to be dealt with in batch processing, especially in complex backgrounds. Future implementations of this method will enable the differentiation between dead and live cells by specific fluorescence.
Acknowledgement:
This work was supported by the National Natural Science Foundation of China (NSFC grants #31900130) and the Fundamental Research Funds for the Central Universities (grants #2019NTST12).
[1] M. Kwolek-Mirek, R. Zadrag-Tecza, Comparison of methods used for assessing the viability and vitality of yeast cells. FEMS Yeast Res 14 (2014) 1068-79.
[2] C. Wilson, R. Lukowicz, S. Merchant, H. Valquier-Flynn, J. Caballero, J. Sandoval, M. Okuom, C. Huber, T.D. Brooks, E. Wilson, B. Clement, C.D. Wentworth, A.E. Holmes, Quantitative and Qualitative Assessment Methods for Biofilm Growth: A Mini-review. Res Rev J Eng Technol 6 (2017).
[3] G.M. Knight, E. Dyakova, S. Mookerjee, F. Davies, E.T. Brannigan, J.A. Otter, A.H. Holmes, Fast and expensive (PCR) or cheap and slow (culture) A mathematical modelling study to explore screening for carbapenem resistance in UK hospitals. BMC Med 16 (2018) 141.
[4] B. Song, B. Zhuge, H. Fang, J. Zhuge, [Analysis of the chromosome ploidy of Candida glycerinogenes]. Wei Sheng Wu Xue Bao 51 (2011) 326-31.
[5] J. Tang, F.U. Qiangqiang, M. Chen, L.I. Chunmei, L.U. Jiazheng, G.P. University, Volume change of H9c2 cells treated with norepinephrine. Shandong Medical Journal (2018).
[6] N. Stolze, C. Bader, C. Henning, J. Mastin, A.E. Holmes, A.L. Sutlief, Automated image analysis with ImageJ of yeast colony forming units from cannabis flowers. J Microbiol Methods 164 (2019) 105681.
[7] Z. Huang, J. Zheng, C. Shi, Q. Chen, Flow cytometry-based method facilitates optimization of PMA treatment condition for PMA-qPCR method. Molecular & Cellular Probes (2018) S0890850818300756.
[8] K.M. Mckinnon, Flow Cytometry: An Overview. Other 120 (2018).
[9] M. Oravcova, M.N. Boddy, Recruitment, loading, and activation of the Smc5-Smc6 SUMO ligase. Curr Genet 65 (2019) 669-676.
[10] C.U. Mårtensson, C. Priesnitz, J. Song, L. Ellenrieder, K.N. Doan, F. Boos, A. Floerchinger, N. Zufall, S. Oeljeklaus, B. Warscheid, T. Becker, Mitochondrial protein translocation-associated degradation. Nature 569 (2019) 679-683.
[11] Y.T. Wang, W.Y. Hsiao, S.W. Wang, The fission yeast Pin1 peptidyl-prolyl isomerase promotes dissociation of Sty1 MAPK from RNA polymerase II and recruits Ssu72 phosphatase to facilitate oxidative stress induced transcription. Nucleic Acids Res (2021).
[12] R.E. Workman, T. Pammi, B.T.K. Nguyen, L.W. Graeff, E. Smith, S.M. Sebald, M.J. Stoltzfus, C.W. Euler, J.W. Modell, A natural single-guide RNA repurposes Cas9 to autoregulate CRISPR-Cas expression. Cell (2021).
[13] X. Zhao, X. Li, P. Zhang, C. Li, W. Feng, X. Zhu, D. Wei, Effects of 5’-3’ Exonuclease Xrn1 on Cell Size, Proliferation and Division, and mRNA Levels of Periodic Genes in Cryptococcus neoformans. Genes (Basel) 11 (2020).
[14] M.Q. Claire, G. Allen, C. Vasiliy, K. Lee, B.A. Cimini, K.W. Karhohs, D. Minh, L. Ding, S.M. Rafelski, T. Derek, CellProfiler 3.0: Next-generation image processing for biology. Plos Biology 16 (2018) e2005970.
[15] D. Houzet, S. Huet, A. Rahman, SysCellC: a data-flow programming model on multi-GPU. Procedia Computer Science 45 (2010).
[16] X. Li, H. Yang, H. Huang, T. Zhu, CELLCOUNTER: Novel open-source software for counting cell migration and invasion in vitro. Journal of Biomedicine and Biotechnology 2014 (2014) 863564.
[17] G. Quentin, M.R. Mh., OpenCFU, a New Free and Open-Source Software to Count Cell Colonies and Other Circular Objects. Plos One 8 (2013) e54072.
[18] I.V. Grishagin, Automatic cell counting with ImageJ. Analytical Biochemistry 473 (2015) 63-65.
[19] S. Berg, D. Kutra, T. Kroeger, C.N. Straehle, B.X. Kausler, C. Haubold, M. Schiegg, J. Ales, T. Beier, M. Rudy, ilastik: interactive machine learning for (bio)image analysis. Nature Methods (2019).
[20] L. Fiaschi, R. Nair, U. Koethe, F.A. Hamprecht, Learning to count with regression forest and structured labels, International Conference on Pattern Recognition, 2012.