IC4R009-Phenomics-2013-23578473 - Revision history

Guangyi: /* Current applications of phenomics in rice or other cereal crops */

2016-07-22T11:02:49Z

‎Current applications of phenomics in rice or other cereal crops

Guangyi: /* Conclusions and future directions */

2016-07-22T11:02:25Z

‎Conclusions and future directions

Guangyi: /* Conclusions and future directions */

2016-07-22T11:01:53Z

‎Conclusions and future directions

Guangyi: /* Research Findings */

2016-07-22T10:57:37Z

‎Research Findings

Guangyi: /* Current applications of phenomics in rice or other cereal crops */

2016-07-22T10:53:44Z

‎Current applications of phenomics in rice or other cereal crops

Guangyi: /* Current applications of phenomics in rice or other cereal crops */

2016-07-22T10:53:15Z

‎Current applications of phenomics in rice or other cereal crops

Guangyi: /* Plant Culture & Treatment */

2016-07-22T10:51:48Z

‎Plant Culture & Treatment

Guangyi: /* Key technologies in plant phenomics */

2016-07-22T10:47:48Z

‎Key technologies in plant phenomics

Guangyi: /* Key technologies in plant phenomics */

2016-07-22T10:46:37Z

‎Key technologies in plant phenomics

Guangyi: /* Key technologies in plant phenomics */

2016-07-22T10:45:59Z

‎Key technologies in plant phenomics

@@ Line 12: / Line 12: @@
 *Both grain production and grain quality decrease as plants are damaged by insects and disease; it is, therefore,important to detect and classify the plant infestations at an early stage [58]. To identify rice blast disease at the seedling stage, a near-infrared hyperspectral imaging system was developed to scan clipped leaves, with an overall accuracy of classification (infected and healthy leaves) of approximately 92% [59]. In addition to these destructive measurement techniques, there are several approaches to achieve real-time and dynamic screening in vivo for pot-grown rice or field-grown cereals. With a color-based corner detection algorithm, visible imagebased methods can detect plant-hopper infestations on the stems of pot-grown rice [60]. Integrating hyper-spectral and fluorescence imaging enables the detection of yellow rust in a winter wheat field, and the overall discrimination performance can reach 99% with a selforganizing map neural network [61].
 *Yield is a complex agronomic trait that is determined by the grain number per plant and grain weight (influenced by grain size). Utilizing a less-expensive flatbed scanner for image acquisition, a user-coded ImageJ software plugin was developed to determine the major orthogonal dimensions of the grains (grain length, grain width) [64]and to analyze the sieveless particle size distribution [65].To accelerate the measurements of spikelet number, a bimodal scanner using visible light imaging and X-ray digital radiography (DR) was employed for the rapid and simultaneous measurement of filled/unfilled rice spikelets [66]. Furthermore, to achieve fully automated yield trait scoring, an integrated facility has been developed to thresh rice panicles, evaluate rice yield traits, and pack filled spikelets. This novel machine vision-based facility is highly accurate (mean absolute percentage error is less than 5%) and highly efficient (1440 plants per continuous 24 hours workday) [20��].
 ==Conclusions and future directions==

@@ Line 14: / Line 14: @@
 ==Conclusions and future directions==
-*Because of the robust genetic technologies, the functional analysis of the rice genome has entered into a highthroughput stage, and the RICE2020 project has been proposed to determine the function of every gene in the rice genome by the year 2020 [77]. To achieve levels of quality and speed comparable to those of genomics,reliable, automatic, multifunctional, and high-throughput phenotyping platforms should be developed using various novel technologies (Figure 1). There is advanced progress in plant phenomics, particularly in Europe (e.g., at the IPK, Leibniz Institute of Plant Genetics and Crop Plant Research) and Australia (at the APPF, Australian Plant Phenomics Facility). However, because rice is a staple food in many developing countries, more efforts to develop low cost and high performance rice phenomic technologies are needed.Besides, with multifunctional phenotyping tools obtaining a large quantity of images and data, how to run the data-storage, handling and analysis will be another challenge in plant phenomics. The data volume mainly depends on the resolution of the imaging detectors and the numbers of acquired image from each inspection. And the data analysis methods, such as principal components analysis (PCA) [78], support vector machine (SVM) [79],and artificial neural network (ANN) [80], are often used for data dimension reduction and efficient parameters extraction. In future, to further promote the application of plant phenotyping, less expensive and sophisticated data analysis infrastructures (e.g., HTPheno [48�] and IAP [81] incorporating the open-source software ImageJ) need to be developed and popularized.Thus, in our opinion, convincing rice scientists to accept or even rely on digital phenotyping platforms, reducing
+*Because of the robust genetic technologies, the functional analysis of the rice genome has entered into a highthroughput stage, and the RICE2020 project has been proposed to determine the function of every gene in the rice genome by the year 2020 [77]. To achieve levels of quality and speed comparable to those of genomics,reliable, automatic, multifunctional, and high-throughput phenotyping platforms should be developed using various novel technologies (Figure 1). There is advanced progress in plant phenomics, particularly in Europe (e.g., at the IPK, Leibniz Institute of Plant Genetics and Crop Plant Research) and Australia (at the APPF, Australian Plant Phenomics Facility). However, because rice is a staple food in many developing countries, more efforts to develop low cost and high performance rice phenomic technologies are needed.Besides, with multifunctional phenotyping tools obtaining a large quantity of images and data, how to run the data-storage, handling and analysis will be another challenge in plant phenomics. The data volume mainly depends on the resolution of the imaging detectors and the numbers of acquired image from each inspection. And the data analysis methods, such as principal components analysis (PCA) [78], support vector machine (SVM) [79],and artificial neural network (ANN) [80], are often used for data dimension reduction and efficient parameters extraction. In future, to further promote the application of plant phenotyping, less expensive and sophisticated data analysis infrastructures (e.g., HTPheno [48�] and IAP [81] incorporating the open-source software ImageJ) need to be developed and popularized.Thus, in our opinion, convincing rice scientists to accept or even rely on digital phenotyping platforms, reducing the platform costs, and developing efficient data storage and analysis infrastructures are the main challenges for the future. However, we are confident that these reliable,high-throughput phenotyping tools will give plant scientists new insights into the information encoded in the rice genome.
 <br>
 [[File:IC4R009-Phenomics-2013-23578473-f1.png |center |thumb |10000px |''''''Fig.1. The main techniques (agronomy, robotics, photonics and computer analyses) needed in plant phenotyping platforms. Using robotics, the rice plants to be screened are transported to the inspection unit. The inspection chamber, which is the core of the phenotyping platform, carries out the noninvasive, high-throughput screening of plant phenotypic traits using photonics and computers. After image analysis, the quantified traits, environmental data(e.g., illumination, temperature, irrigation, fertilizer) and genotype are all managed in a database, which produces a ‘phenotype–genotype model’ and allows the simulation or predication of responses for special genotypes in different environmental scenarios.'''''']]

@@ Line 14: / Line 14: @@
 ==Conclusions and future directions==
-*Because of the robust genetic technologies, the functional analysis of the rice genome has entered into a highthroughput stage, and the RICE2020 project has been proposed to determine the function of every gene in the rice genome by the year 2020 [77]. To achieve levels of quality and speed comparable to those of genomics,reliable, automatic, multifunctional, and high-throughput phenotyping platforms should be developed using various novel technologies (Figure 1). There is advanced progress in plant phenomics, particularly in Europe (e.g., at the IPK, Leibniz Institute of Plant Genetics and Crop Plant Research) and Australia (at the APPF, Australian Plant Phenomics Facility). However, because rice is a staple food in many developing countries, more efforts to develop low cost and high performance rice phenomic technologies are needed.Besides, with multifunctional phenotyping tools obtaining a large quantity of images and data, how to run the data-storage, handling and analysis will be another challenge in plant phenomics. The data volume mainly depends on the resolution of the imaging detectors and the numbers of acquired image from each inspection. And the data analysis methods, such as principal components analysis (PCA) [78], support vector machine (SVM) [79],and artificial neural network (ANN) [80], are often used for data dimension reduction and efficient parameters
+*Because of the robust genetic technologies, the functional analysis of the rice genome has entered into a highthroughput stage, and the RICE2020 project has been proposed to determine the function of every gene in the rice genome by the year 2020 [77]. To achieve levels of quality and speed comparable to those of genomics,reliable, automatic, multifunctional, and high-throughput phenotyping platforms should be developed using various novel technologies (Figure 1). There is advanced progress in plant phenomics, particularly in Europe (e.g., at the IPK, Leibniz Institute of Plant Genetics and Crop Plant Research) and Australia (at the APPF, Australian Plant Phenomics Facility). However, because rice is a staple food in many developing countries, more efforts to develop low cost and high performance rice phenomic technologies are needed.Besides, with multifunctional phenotyping tools obtaining a large quantity of images and data, how to run the data-storage, handling and analysis will be another challenge in plant phenomics. The data volume mainly depends on the resolution of the imaging detectors and the numbers of acquired image from each inspection. And the data analysis methods, such as principal components analysis (PCA) [78], support vector machine (SVM) [79],and artificial neural network (ANN) [80], are often used for data dimension reduction and efficient parameters extraction. In future, to further promote the application of plant phenotyping, less expensive and sophisticated data analysis infrastructures (e.g., HTPheno [48�] and IAP [81] incorporating the open-source software ImageJ) need to be developed and popularized.Thus, in our opinion, convincing rice scientists to accept or even rely on digital phenotyping platforms, reducing
 the platform costs, and developing efficient data storage and analysis infrastructures are the main challenges for the future. However, we are confident that these reliable,high-throughput phenotyping tools will give plant scientists new insights into the information encoded in the rice genome.
 <br>
-[[File:IC4R009-Phenomics-2013-23578473.png |center |thumb |10000px |'''''']]
+[[File:IC4R009-Phenomics-2013-23578473-f1.png |center |thumb |10000px |''''''Fig.1. The main techniques (agronomy, robotics, photonics and computer analyses) needed in plant phenotyping platforms. Using robotics, the rice plants to be screened are transported to the inspection unit. The inspection chamber, which is the core of the phenotyping platform, carries out the noninvasive, high-throughput screening of plant phenotypic traits using photonics and computers. After image analysis, the quantified traits, environmental data(e.g., illumination, temperature, irrigation, fertilizer) and genotype are all managed in a database, which produces a ‘phenotype–genotype model’ and allows the simulation or predication of responses for special genotypes in different environmental scenarios.'''''']]
 <br>

@@ Line 13: / Line 13: @@
 *Yield is a complex agronomic trait that is determined by the grain number per plant and grain weight (influenced by grain size). Utilizing a less-expensive flatbed scanner for image acquisition, a user-coded ImageJ software plugin was developed to determine the major orthogonal dimensions of the grains (grain length, grain width) [64]and to analyze the sieveless particle size distribution [65].To accelerate the measurements of spikelet number, a bimodal scanner using visible light imaging and X-ray digital radiography (DR) was employed for the rapid and simultaneous measurement of filled/unfilled rice spikelets [66]. Furthermore, to achieve fully automated yield trait scoring, an integrated facility has been developed to thresh rice panicles, evaluate rice yield traits, and pack filled spikelets. This novel machine vision-based facility is highly accurate (mean absolute percentage error is less than 5%) and highly efficient (1440 plants per continuous 24 hours workday) [20��].
-==Research Findings==
+==Conclusions and future directions==
 ==Labs working on this Project==

@@ Line 11: / Line 11: @@
 *Using the plant age and plant area calculated by the Scanalyzer 3D, a modified model enables the more accurate high-throughput estimation of biomass for cereal plants under saline conditions [19��]. IR thermal imaging is also commonly used to quantify the osmotic stress response to salinity or drought in cereal crops [55].By precisely controlling the environments of each potgrown cereal plant in the greenhouse, these accurate,high-throughput phenotyping tools can overcome the limitations of current salinity-resistance or drought-resistance research.
 *Both grain production and grain quality decrease as plants are damaged by insects and disease; it is, therefore,important to detect and classify the plant infestations at an early stage [58]. To identify rice blast disease at the seedling stage, a near-infrared hyperspectral imaging system was developed to scan clipped leaves, with an overall accuracy of classification (infected and healthy leaves) of approximately 92% [59]. In addition to these destructive measurement techniques, there are several approaches to achieve real-time and dynamic screening in vivo for pot-grown rice or field-grown cereals. With a color-based corner detection algorithm, visible imagebased methods can detect plant-hopper infestations on the stems of pot-grown rice [60]. Integrating hyper-spectral and fluorescence imaging enables the detection of yellow rust in a winter wheat field, and the overall discrimination performance can reach 99% with a selforganizing map neural network [61].
-*Yield is a complex agronomic trait that is determined by the grain number per plant and grain weight (influenced by grain size). Utilizing a less-expensive flatbed scanner for image acquisition, a user-coded ImageJ software plugin was developed to determine the major orthogonal dimensions of the grains (grain length, grain width) [64]and to analyze the sieveless particle size distribution [65].To accelerate the measurements of spikelet number, a bimodal scanner using visible light imaging and X-ray
+*Yield is a complex agronomic trait that is determined by the grain number per plant and grain weight (influenced by grain size). Utilizing a less-expensive flatbed scanner for image acquisition, a user-coded ImageJ software plugin was developed to determine the major orthogonal dimensions of the grains (grain length, grain width) [64]and to analyze the sieveless particle size distribution [65].To accelerate the measurements of spikelet number, a bimodal scanner using visible light imaging and X-ray digital radiography (DR) was employed for the rapid and simultaneous measurement of filled/unfilled rice spikelets [66]. Furthermore, to achieve fully automated yield trait scoring, an integrated facility has been developed to thresh rice panicles, evaluate rice yield traits, and pack filled spikelets. This novel machine vision-based facility is highly accurate (mean absolute percentage error is less than 5%) and highly efficient (1440 plants per continuous 24 hours workday) [20��].
 ==Research Findings==

@@ Line 8: / Line 8: @@
 <br>
-==Plant Culture & Treatment==
+==Current applications of phenomics in rice or other cereal crops==
 ==Research Findings==

@@ Line 4: / Line 4: @@
 ==Key technologies in plant phenomics==
 *Since the first digital camera was invented by Eastman Kodak in 1975 (www.letsgodigital.org/en/16859/ce-hallof-fame), visible light imaging technology has been widely adopted in plant science due to its low cost and ease of maintenance. With a similar wavelength (400–700 nm) perception as the human eye, two-dimensional(2D) photography can be used to analyze shoot biomass[18,19��], yield-related traits [20��], leaf morphology [23],panicle traits [24], and the system architecture traits of washed roots or roots grown in transparent media [25].
-*Because of internal molecular movements, all objects emit characteristic infrared radiation [29]. Two popular infrared imaging devices can be used to screen radiation images: a near-infrared (NIR, wavelength of approximately 0.9–1.7 mm) imaging device and a far-infrared(Far-IR, wavelength of approximately 7.5–13.5 mm) imaging device. Healthy green plants reflect a large proportion of NIR light from 800 to 1400 nm, whereas the soil reflects little NIR light; moreover, soil and unhealthy plants reflect considerably more red wavelength light as compared with healthy plants. For these reasons, many studies have combined NIR imaging and visible imaging to detect vegetative indices. A Crop Phenology Recording System (CPRS) has been developed for monitoring rice growth. CPRS uses visible light imaging to derive the visible atmospherically resistant index and uses nearinfrared imaging (830 nm) to derive the night-time relative brightness index and then establishes the relationship between the camera-derived indices and
+*Because of internal molecular movements, all objects emit characteristic infrared radiation [29]. Two popular infrared imaging devices can be used to screen radiation images: a near-infrared (NIR, wavelength of approximately 0.9–1.7 mm) imaging device and a far-infrared(Far-IR, wavelength of approximately 7.5–13.5 mm) imaging device. Healthy green plants reflect a large proportion of NIR light from 800 to 1400 nm, whereas the soil reflects little NIR light; moreover, soil and unhealthy plants reflect considerably more red wavelength light as compared with healthy plants. For these reasons, many studies have combined NIR imaging and visible imaging to detect vegetative indices. A Crop Phenology Recording System (CPRS) has been developed for monitoring rice growth. CPRS uses visible light imaging to derive the visible atmospherically resistant index and uses nearinfrared imaging (830 nm) to derive the night-time relative brightness index and then establishes the relationship between the camera-derived indices and the agronomic traits [30].
 *In recent years, several modern optical imaging techniques, for instance, 3D structural tomography and functional imaging, have been developed and expanded to improve living plant visualization. The rice plants serve as ‘patients’ in a novel use of X-ray computed tomography(CT) scanners to estimate the tiller number [38��].Equipped with an acceleration algorithm using the adaptive minimum enclosing rectangle (AMER) and graphics processing unit (GPU), the entire tiller inspection time of one plant is less than 200 ms [39]. Moreover, the incorporation of various optical sensors, such as visible and infrared digital cameras, provides this system with the potential to achieve the advanced screening of multiple traits for pot-grown rice plants within one chamber [38��].
 <br>