Facial Recognition Technology for Automated Stomatal Aperture Measurement

DeepStomata: Facial Recognition Technology for Automated Stomatal Aperture Measurement

Toda Y., Toh S., Bourdais G., Robatzek S., Macleabn D., Kinoshita T. (2018)

Yosuke Toda, Shigeo Toh, Gildas Bourdais, Silke Robatzek, Dan Maclean, Toshinori Kinoshita,

In bioRxiv, 365098 – https://doi.org/10.1101/365098

https://www.biorxiv.org/content/10.1101/365098v1

Workflow and processing accuracy of automatic stomatal aperture quantification.a, Schematic diagram of facial detection. First, the sub-region of an image that contains a face is detected and cropped (Face Detection) using predefined features or machine learning. Next, the cropped images are passed to the trained convolutional neural networks (CNNs) to infer complex factors such as age, biological sex, emotions, or identity (name). b, Schematic diagram of our proposed method. First, the subregions that contain stomata are obtained by the HOG detector (Step 1; See Supplementary Figure 1a for details of the constructed HOG detector). Next, cropped images are processed by the CNN to infer whether the respective input image is an open, partially open, or closed stoma, or a false positive that does not contain a stoma (Step 2; See Supplementary Figure 2a for details of the CNN architecture). Images classified as false positives are discarded from the analysis. Closed stomata are assigned a 0-µm stomatal aperture. Open and partially open stomata are passed to the pore quantification step (Step 3), in which the images are processed to create and label their segmented regions. To discriminate the stomatal pore, manually set criteria were introduced to filter out non-pore regions (See Methods for details). c, Scatter plot of automatically quantified stomatal apertures versus manually quantified apertures. The classifications denoted by the CNN are displayed in different colours. The equation and R-squared value of the regression line are displayed. Data categorized by classification are summarized in dd, Box plots showing the stomatal apertures of the closed, partially open, and open stomata classified by the CNN. Statistical differences were determined using a one-way ANOVA followed by a Tukey post-hoc analysis; different letters indicate significant differences (p < 0.01).

Abstract

Stomata are an attractive model for studying the physiological responses of plants to various environmental stimuli13. Of the morphological parameters that represent the degree of stomatal opening, the length of the minor axis of the stomatal pore (the stomatal aperture) has been most commonly used to dissect the molecular basis of its regulation. Measuring stomatal apertures is time consuming and labour intensive, preventing their use in large-scale studies. Here, we completely automated this process by developing a program called DeepStomata, which combines stomatal region detection and pore isolation by image segmentation. The former, which comprises histograms of oriented gradients (HOG)-based stomatal detection and the convolutional neural network (CNN)-based classification of open/closed-state stomata, acts as an efficient conditional branch in the workflow to selectively quantify the pores of open stomata. An analysis of batches of images showed that the accuracy of our automated aperture measurements was equivalent to that of manual measurements, however had higher sensitivity (i,e., lower false negative rate) and the process speed was at least 80 times faster. The outstanding performance of our proposed method for automating a laborious and repetitive task will allow researchers to focus on deciphering complex phenomena.

The utilization of foliar micromorphological characters for the identification of fodder grass taxa

PLATE 2(A) Foliar epidermis SEM view (a)Agrostisgigantea(b)Avena sativa(c)Bromusjaponicus(d)Dactylisglomerata(e)Loliumtemulentum(f)Phalaris minor(g)Poaannua(h)Poainfirma(i)Polypogonmonspeliensis(j)Arundodonax(k)Phragmitesaustralis(l)Aristidaadscensionis(m)Acrachneracemosa(n)Cynodondactylon(o)Dactylocteniumaegyptium(p)Desmostachyabipinnata(q)Eleusineindica(r)Enneapogonpersicus(s)Eragrostis japonica(t)Eragrostis minor(u)Eragrostispilosa(v)Leptochloapanicea(w)Tetrapogonvillosus(x)Apludamutica(y)Bothriochloabladhii.(B) Foliar epidermis SEM view (a)Brachiariaramosa(b)Brachiariareptans(c)Cenchrusbiflorus(d)Cenchrusciliaris(e)Cenchruspennisetiformis(f)Cenchrussetiger(g)Chrysopogonaucheri(h)Chrysopogonzizanioides(i)Cymbopogonjwarancusa(j)Dichanthiumannulatum(k)Digitariaciliaris(l)Digitarialongiflora(m)Echinochloacolona(n)Echinochloa crus-galli(o)Heteropogoncontortus(p)Imperatacylindrica(q)Ottochloacompressa(r)Panicumantidotale(s)Paspalidiumdistichum(t)Pennisetumorientale(u)Saccharum bengalense(v)Saccharumspontaneum(w)Setariapumila(x)Setariaverticillata(y)Setariaviridis. (C) Foliar epidermis SEM view (a)Sorghum bicolor(b)Sorghum halepense(c)Zea mays
(3) (PDF) Light and scanning electron microscopy-based foliar micro morphological tools for the identification of fodder grass taxa. Available from: https://www.researchgate.net/publication/341203653_Light_and_scanning_electron_microscopy-based_foliar_micro_morphological_tools_for_the_identification_of_fodder_grass_taxa [accessed May 13 2020].

Light and scanning electron microscopy-based foliar micro-morphological tools for the identification of fodder grass taxa

by Harun N., Shaheen S., Ahmad M., Shahid M. N. (2020)

Nidaa Harun, Shabnum Shaheen, Mushtaq Ahmad, Muhammad Naveed Shahid,

In Microscopy Research and Technique – DOI: 10.1002/jemt.23490

https://www.researchgate.net/publication/341203653_Light_and_scanning_electron_microscopy-based_foliar_micro_morphological_tools_for_the_identification_of_fodder_grass_taxa

Abstract

Fertile plain of Central Punjab Pakistan is rich with fodder grasses and from centuries the local inhabitants of this area have been using their regional grasses for ruminant feeding. However, they always faced difficulties in identification because of their overlapping vernacular names, more or less identical leaf shapes, indefinite variations in stem branching pattern, and reduced floral parts. Hence, the current study has provided a detailed and comprehensive micro‐morphological analysis of 53 ethnobotanical fodder grass taxa. A variety of quantitative and qualitative leaf epidermal micromorphological traits was studied and results reported epidermal characters like stomatal index, silica bodies, prickles, microhairs, hook cells as most diagnostic in delimitation of species, and genera. As stomatal index was 79% in Poa annua while in its closely related species Poa infirmai was 85%. Similarly, Cenchrus ciliaris can be differentiated from Cenchrus pennisetiformis on the basis of silica body shape as butterfly shaped in former and dumbbell shaped in later one. Moreover, prickles were present in Chrysopogon aucheri while absent in Chrysopogon zizanioides. Hence, overall this study declared that diversity and variations in foliar micromorphological characters are valuable and supportive in the identification of grasses at the specific and generic level. • The current research validates the utilization of foliar micromorphological characters for the identification of fodder grass taxa. Some of valuable foliar micromorphological characters like stomatal index, silica bodies, prickles, microhairs, and hook cells were regarded as taxonomically significant.

Method: the stomatal movement is inferred by simple monitoring of the fluorescence intensity in the nucleus of the stomata

Hoechst-tagged Fluorescein Diacetate for the Fluorescence Imaging-based Assessment of Stomatal Dynamics in Arabidopsis thaliana

by Takaoka Y., Miyagawa S., Nakamura A., Egoshi S.,Tsukiji S., Ueda M. (2020)

Yousuke Takaoka, Saki Miyagawa, Akinobu Nakamura, Syusuke Egoshi, Shinya Tsukiji, Minoru Ueda,

===

In Sci Rep 105333 – https://doi.org/10.1038/s41598-020-62239-w

https://www.nature.com/articles/s41598-020-62239-w#citeas

(a,b) Differential interference images (DIC) and fluorescent (tdTomato or Fluorescein) microscopic images of HoeAc2Fl-stained stomata of P35S ::H2B-tdTomato35S::H2B-tdTomato in the dark (a) or light (b) conditions; only the closed stomata were stained with HoeAc2Fl (the stomatal aperture was 2.15 µm), whereas opened stomata were not (the stomatal aperture was 4.70 µm) (see images in the fluorescein channel). The scale bars, 10 µm. (c) Relationship between stomatal apertures and fluorescent intensity of the

Abstract

In plants, stomata regulate water loss through transpiration for plant growth and survival in response to various environmental stressors; and simple methods to assess stomatal dynamics are needed for physiological studies. Herein, we report a fluorescence-imaging-based method using fluorescein diacetate tagged with Hoechst 33342, a nuclear staining chemical probe (HoeAc2Fl) for the qualitative assessment of stomatal dynamics. In our method, the stomatal movement is inferred by simple monitoring of the fluorescence intensity in the nucleus of the stomata.

Conclusion

HoeAc2Fl is proposed as a tool to easily and quickly assess whether plant stomata are open or closed based on its selectivity for the guard cells of closed stomata. The mechanistic basis for this selectivity is unknown. When the stomata were stained by HoeAc2Fl, the fluorescence was observed only from closed stomata. The clear threshold of the fluorescence provides objective criteria for the assessment of stomatal dynamics, although it is not quantitative. Instant determination of stomatal dynamics by measuring the fluorescence of HoeAc2Fl with objective analyses is expected to enable high-throughput screening of chemical libraries, which may lead to the discovery of novel chemical probes that can improve our understanding of plant responses to changes in their environments, and ultimately lead to improved crop production.

StomataCounter technology

Fig. 1 Architecture of the deep convolutional
neural network (DCNN) and classification
tasks. Left: training and testing procedure.
First column: target patches were extracted
and centered around human-labeled stomata
center positions; distractor patches were
extracted in all other regions. A binary image
classification network was trained. Second
column: The image classification network
was applied fully convolutional to the test
image to produce a prediction heatmap. On
the thresholded heatmap, peaks were
detected and counted.

StomataCounter: a neural network for automatic stomata
identification and counting

by Fetter K. C., Eberhardt S., Barclay R. S., Wing S., Keller S. R. (2019)

Karl C. Fetter1,2* , Sven Eberhardt3, Rich S. Barclay2 , Scott Wing2 and Stephen R. Keller1


1 Department of Plant Biology, University of Vermont, Burlington, VT 05405, USA;

2 Department of Paleobiology, Smithsonian Institution, National Museum of Natural History, Washington,
DC 20560, USA;

3 Amazon.com, Inc., Seattle, WA 98121, USA

===

In New Phytologist – doi: 10.1111/nph.15892 –

http://www.uvm.edu/~kellrlab/LabManuscripts/Fetter_etal_2019.pdf

Fig. 2 Sample sizes of images for the top 20 families represented in the training (yellow) and test (blue) datasets (a). Examples of pre- and post-analysis
images. A probability heatmap map is overlain onto the input image in the red channel. Detected stomata marked with circles with peak values given in
green (b). Species in image pairs: Adiantum peruvianum, Begonia loranthoides (top row); Chaemaerauthum gadacardii, Echeandia texensis (second row);
Ginkgo biloba, Populus balsamifera (third row); Trillium luteum, Pilea libanensis (bottom row). Bars: (top and second rows) 20 lm, (row three) 25 lm;
(bottom row) 50 lm.

Summary

Stomata regulate important physiological processes in plants and are often phenotyped by researchers in diverse fields of plant biology. Currently, there are no user-friendly, fully automated methods to perform the task of identifying and counting stomata, and stomata density is generally estimated by manually counting stomata.
We introduce StomataCounter, an automated stomata counting system using a deep convolutional neural network to identify stomata in a variety of different microscopic images. We use a human-in-the-loop approach to train and refine a neural network on a taxonomically
diverse collection of microscopic images.
Our network achieves 98.1% identification accuracy on Ginkgo scanning electron microscropy micrographs, and 94.2% transfer accuracy when tested on untrained species.
To facilitate adoption of the method, we provide the method in a publicly available website at http://www.stomata.science/.

A New and Quick Method to Count Stomata

Stomata Imprints: A New and Quick Method to Count Stomata and Epidermis Cells

by Meister M. H., Nordenkampf H. R. B. (2001)

M. H. Meister, H. R. Bolhàr Nordenkampf,

In: Reigosa Roger M.J. (eds) Handbook of Plant Ecophysiology Techniques. Springer, Dordrecht – https://doi.org/10.1007/0-306-48057-3_17

https://link.springer.com/chapter/10.1007/0-306-48057-3_17#citeas

Abstract

Comparing several techniques for creating epidermal replicas, we found that imprints on cellulose-di-acetate and on polymethyl-metacrylate can be easily performed within a few seconds, which proves to be a considerable advantage especially in field trials.

The result is a permanent impression of the epidermis’ surface, perfect for long-term storage. In the case of extremely sunken stomata, pleated leaf surfaces, or coniferous needles, we additionally used a cyanacrylate adhesive. Reliable and reproducible results could be achieved for use when analysing the imprints by a drawing microscope or an image analyses program.

The investigation of transpiration rate of leaves as controlled by stomatal aperture

Experimental Studies of the Factors Controlling Transpiration: I. APPARATUS AND EXPERIMENTAL TECHNIQUE

by Gregory F. G., Milthorpe F. L., Pearse H. L., Spencer H. J. (1950)

===

In Journal of Experimental Botany 1(1): 1–14 – https://doi.org/10.1093/jxb/1.1.1

https://academic.oup.com/jxb/article-abstract/1/1/1/653268?redirectedFrom=fulltext

Abstract

Apparatus and experimental techniques are discussed for use in the investigation of transpiration rate of leaves as controlled by stomatal aperture and leaf water content.

The leaf chambers used and the methods adopted for the estimation of the water transpired are described.

The designs of the porometer cups used for the different types of leaves ( Pelargonium and wheat) employed in the work are described. To obviate the difficulty that stomata within the cup behave abnormally, the design employed makes possible a removal of the cup from the leaf except during the short periods required to estimate stomatal resistance to air flow at intervals during the course of an experiment.

In these experiments the water content of the leaf is changed at will by interrupting the water-supply and re-establishing it to permit recovery from wilting. The methods used to carry out this cycle of operations are fully dealt with. Determinations of the transpiration and absorption rates during the experiment and of the final leaf water content make it possible to follow changes in leaf water content throughout the experiment.

An account is given of the methods used for varying the speed of flow, the humidity, and the CO 2 concentration of the air streams.

Readings were closely correlated with measured stomatal apertures

Automated system for following stomatal behavior of plants in growth cabinets

by Allaway W. G., Mansfield T. A. (1969)

In Canadian Journal of Botany 47(12): 1995-1998 – https://doi.org/10.1139/b69-292

https://www.nrcresearchpress.com/doi/abs/10.1139/b69-292?mobileUi=0

Abstract

An automatic porometer recording from eight leaves is described. The apparatus has enabled experiments lasting several days to be performed, with porometer readings every half-hour from each leaf. The porometer cups apparently did not alter the behavior of stomata even after long attachment, and readings were closely correlated with measured stomatal apertures on the upper surface of amphistomatous leaves.

Measurements of stomatal density and stomatal index on leaf/plant surfaces

Measurements of stomatal density and stomatal index on leaf/plant surfaces

by Paul V., Sharma L., Pandey R., Meena R. C. (2017)

Vijay Paul, Laxmi Sharma, Rakesh Panday, R. C. Meena,

In Manual of ICAR sponsored training programme on “Physiological Techniques to Analyze the Impact of Climate change on crop plants” 16–25 January – Division of Plant Physiology IARI, New Delhi, India – pp. 27-30 –

https://www.researchgate.net/profile/Vijay_Paul2/publication/321274422_Physiological_Techniques_to_Analyze_the_Impact_of_Climate_Change_on_Crop_Plants/links/5a181f9e4585155c26a7c6ed/Physiological-Techniques-to-Analyze-the-Impact-of-Climate-Change-on-Crop-Plants.pdf#page=33

Leaf Sensor for Stomata Transpiration Monitoring

Leaf Sensor for Stomata Transpiration Monitoring Using Temperature and Humidity

Pipitsunthonsan P., Sopharat J., Sirisuk P., Chongcheawchamnan M. (2018)

Pronthep Pipitsunthonsan, Jessada Sopharat, Phaophak Sirisuk, Mitchai Chongcheawchamnan,

Pronthep Pipitsunthonsan: Department of Electrical Engineering, Faculty of Engineering, Prince of Songkla University, Songkhla, Thailand

Jessada SopharatDepartment of Earth Science, Faculty of Natural Resources, Prince of Songkla University, Songkhla, Thailand

Phaophak Sirisuk: International College, King Mongkut’s Institute of Technology Ladkrabang, Bangkok, Thailand

Mitchai Chongcheawchamnan: Department of Electrical Engineering, Faculty of Engineering, Prince of Songkla University, Songkhla, Thailand

===

In 21st International Symposium on Wireless Personal Multimedia Communications (WPMC)DOI: 10.1109/WPMC.2018.8713096

https://ieeexplore.ieee.org/document/8713096

Abstract

This paper presents a leaf sensor which was designed for stomatal transpiration detection. The sensor consists of a thermocouple, a humidity sensor and a programmed microcontroller. The thermocouple measures temperature difference between the leaf and its surrounding atmosphere while relative humidity of the atmosphere was recorded. The sensor design was verified with the standard tool and the image of stomata captured with a high-gain camera lens. From the experimental results, data collected from the developed sensor indicates the stomatal status. This proves the potentiality of the proposed sensor for stomatal transpiration detection.

In situ Stomatal Counting by Fluorescence Microscopy and Combined Image Analysis

Epicuticular Phenolics Over Guard Cells: Exploitation for in situ Stomatal Counting by Fluorescence Microscopy and Combined Image Analysis

by Karabourniotis G., Tzobanoglou D., Nikolopoulos D., Liakopoulos G. (2001)

In Annals of Botany, 87(5): 631–639 – https://doi.org/10.1006/anbo.2001.1386

https://academic.oup.com/aob/article/87/5/631/2588449/Abstract

Abstract

Guard cells emit an alkali-induced, blue fluorescence upon excitation by ultraviolet radiation (emission maximum energy at 365 nm). Fluorescence emission of guard cells was brighter than that of the neighbouring epidermal cells in a number of wild and cultivated plants including conifers, but the relative fluorescence intensity and quality was species-dependent. Three representative plants possessing stomatal complexes which differed morphologically were studied: Olea europaea , Vicia faba and Triticum aestivum . Immersing leaves of these plants in chloroform for 30 s (thereby removing epicuticular waxes) significantly reduced the intensity of the fluorescence emitted by guard cells. This indicates that guard cell fluorescence could be due to either an increased concentration of fluorescing compounds (probably wax-bound phenolics), or a thicker cuticular layer covering the guard cells. Given that the alkali-induced blue fluorescence of the guard cells is a common characteristic of all plants examined, it could be used as a rapid and convenient method for in situ measurements of the number, distribution and size of stomatal complexes. The proposed experimental procedure includes a single coating of the leaf surface by, or immersion of the whole leaf in, a 10% solution of KOH for 2 min, washing with distilled water, and direct observation of the leaf surface under the fluorescence microscope. Fluorescence images were suitable for digital image analysis and methodology was developed for stomatal counting using Olea europaea as a model species.