The Center for Plant Systems Biology has several imaging systems and platforms to automatically measure and follow plant growth.
The Weighing, Imaging and Watering Machine (WIWAM) is an imaging platform developed by SMO and VIB for high-throughput phenotyping of Arabidopsis plants. The robot is capable of simultaneously handling a large number of plants and measuring a variety of plant growth parameters at regular time intervals. The standard platform consists of a table with a capacity of 396 pots, and a robotic arm, which regularly moves individual pots to different stations for imaging, weighing and watering (Figure 1). Phenotyping is performed by an image analysis script that extracts rosette growth features, such as the compactness, perimeter and stockiness, of each plant automatically (Figure 2). The image sequences allow constructing reliable growth curves over time. WIWAM replaces labor-intensive manual work, saving time and costs. Because the growth conditions are standardized and larger sets of plants can be screened, the robot also allows obtaining statistically more significant results. In addition, the image analysis provides information which is overlooked by the naked eye. In our department, the WIWAM is mainly used to investigate rosette growth under mild drought conditions, but it can be applied to a whole range of biological questions. Currently, we are looking into the possibility to commercialize this plant phenotyping robot. Furthermore, the WIWAM setup can be tailored to meet your specific needs.
For further information, contact:
Tel:+32 (0)9 331 39 53
Figure 1. WIWAM. (A) Robotic arm and table containing 228 pots. (B) Imaging, weighing, and watering station from top to bottom.
Figure 2. Image analysis of Arabidopsis rosettes grown in soil. These panels illustrate the imaging procedures performed on in vivo-grown Arabidopsis plants. The image analysis algorithm extracts a rosette from its background and measures projected rosette area, perimeter and convex hull. The final image is saved for quick visual control.
Skirycz, A., Vandenbroucke, K., Clauw, P., Maleux, K., De Meyer, B., Dhondt, S., Pucci, A., Gonzalez, N., Hoeberichts, F., Tognetti V.B., Galbiati, M., Tonelli, C., Van Breusegem, F., Vuylsteke, M., Inzé, D. (2011) Survival and growth of Arabidopsis plants given limited water are not equal. Nat. Biotechnol. 29(3):212-4.
In order to monitor growth parameters of Arabidopsis plants in vitro, the In Vitro Growth Imaging System (IGIS) was built. The basic setup consists of a rotating metal disk, which can accommodate up to ten Petri dishes, and a single-lens reflex camera, which captures a top view picture of the germinating seedlings (Figure 3). The disk is put into motion by a step motor driving a central axle. The motor is connected to a programmable logic controller (PLC), allowing the integration of timers and sensors that control the motor. To prevent condensation at the top of the Petri dishes, the disk has to be cooled. Therefore, the system is connected to an air cooling system, propelling cooled air into a metal ring underneath the disk. The phenotyping platform can be used under continuous light conditions, but to enable viewing the plants during the dark period, the platform is equipped with infrared (IR) light emitting diodes (LEDs) with a wavelength peak around 940 nm, which is well outside the visible spectrum for plants. Furthermore, to allow the reflex camera to capture the IR light and to ensure that the pictures taken during the light period are comparable to those taken during the dark period, some adjustments are employed. IGIS is placed in the tissue culture rooms to ensure controlled growth conditions.
During a growth experiment, the system runs for a period of 21 days with the specifications to move one position every six minutes, bringing each of the ten Petri dishes underneath the camera position within the time frame of one hour. For each Petri dish, twelve seeds are evenly spread over the plate, to prevent plants overlapping at the final stages, meaning that the individual rosette growth of 120 plants is followed over 480 hours in a typical time-lapse sequence. An automated image analysis and visualization pipeline was created for IGIS (Figure 4). First, positions of the plants to be analyzed are marked in a so-called mask file (Figure 4C). The selected plants on each plate are numbered automatically based on the position of the corresponding disks in the mask images (Figure 4D). Then, individual plants are extracted and rosette parameters are measured. These values are extracted for all plants on every plate at each time point and are used for further calculation of mean rosette area, compactness values and other time-derived parameters, like relative rosette growth rate and relative change in rosette compactness. Finally, the data is automatically plotted in a number of graphs, revealing temporal rosette growth patterns of in vitro-grown Arabidopsis plants (Figure 4G-I).
Figure 3. Picture of the IGIS setup to monitor in vitro Arabidopsis rosette growth in day/night conditions using an IR camera. The major parts of imaging platform are indicated.
Figure 4. Image and data analysis of pictures captured with IGIS. For details see text. (A) Representative picture of a Petri dish captured with a camera adjusted for IR imaging. (B) Representative picture of a Petri dish captured with a normal single-lens reflex camera. Red circles indicate non-germinated seeds and badly grown plants which should be excluded from the analysis. (C) A mask image containing black disks on the position of selected plants. (D) Automated numbering of the plants based on the position of the disks in the mask image. (E) Selection of the blue channel after an RGB split of the original color figure in (B) to increase the contrast between the plants and their background. (F) Extraction of the rosettes after the application of a threshold and a reconstructive opening based on the disks in the mask image. (G-I) Automatically generated graphs after image and data analysis of the complete time-lapse, showing data of wild-type and AVP1-overexpressing plants grown on different growth media under day/night light conditions. (G) Rosette area during development. The inset shows the same data with log scale. Error bars show SE. (H) Rosette compactness during development. The inset shows rosette compactness between day 14 and 15. Error bars show SE. (I) Relative rosette growth rate per hour during development.
Another phenotyping tool, the Multi-camera In vivo Rosette Growth Imaging System (MIRGIS), was developed in the Systems Biology and Yield group. It is well known that the physical environment immensely affects plant phenotypes, and that growing Arabidopsis plants in soil can lead to increased biological variation within identical genotypes. This new platform allows automatic tracking of individual plants after germination. The system consists of several single-lens reflex cameras and a laptop. Three cameras are mounted above six trays of Arabidopsis plants in a custom-designed rack, together with fluorescent lamps, and placed in a controlled growth chamber (Figure 5). An in-house computer script has been developed so that each camera would take a picture of the two trays underneath on a daily basis. The setup can image a total of 144 plants, per three cameras. The basic concept is that four seeds are sown per pot and followed from early development. At a specific point during development, a seedling selection is done in order to reduce the variation within the pool of in soil-grown plants, which mainly arises by differences in germination time. This is performed before the plants start to overlap. Therefore, the projected areas of all seedlings of a specific genotype (which are usually spread over the different trays) are measured and the median seedling area of that genotype is calculated. Then, the seedling with an area closest to the median value is selected within each pot of that genotype. A selection image is generated in which the selected plant is indicated by a red circle. This image is used to manually remove the remaining plants in the pots. This methodology ensures that the seedling that is selected per pot is the one closest to the average plant in the population of its genotype. The location of the selected seedling is afterwards used to extract the plants throughout the complete image sequence, allowing for the construction of growth curves and the calculation of relative growth rates. This phenotyping setup is designed for Arabidopsis rosette growth analysis.
Figure 5. MIRGIS. (A) Six trays with 144 pots standing on a custom-designed rack in a growth room. (B) Three identical single-lens reflex cameras (Canon EOS 450D) fixed at the top of the rack between the fluorescent lamps connected to a single laptop.
The maize WIWAM, also called 'SHRIMPY' (System for High-throughput Recording and Imaging of Maize Phenotypes related to Yield), is an imaging robot designed for crops in the Systems Biology of Yield group of Professor Dirk Inzé. The system has a capacity of 156 plants. Like the Arabidopsis WIWAM, it is located in a controlled-environment growth chamber and enables automated imaging and automated weighing and irrigation of plants according to a preset scheme, specific for each plant or group of plants. This may include fixed quantities of solution, or irrigation to a certain target weight determined by the requested soil humidity levels. The system is thus ideally suited to impose soil water deficit treatments. In contrast to the Arabidopsis WIWAM, plants are positioned in tables which are sorted in rows. The robot arm pushes the tables aside to create the required space in between the rows for the robot arm to locate a pot and lift it out of the table. The robot then takes it to the rotating platform of the weighing and irrigation station. For the imaging of plants, pots are taken to a designated area at the back of the growth chamber where plants are rotated in front of an RGB camera for multiple angle imaging. The acquired images are used for the three-dimensional reconstruction of plants and the extraction of quantitative growth-related traits. Experimental setups, experiment metadata and results are managed by PIPPA (PSB Interface for Plant Phenotype Analysis), the central user interface and database.
The maize WIWAM was developed and constructed in collaboration with SMO.
PHENOVISION is a greenhouse infrastructure for automated, high-throughput phenotyping of crops with a capacity of 392 plants. Pots are transported in carriers on a conveyor belt system. Both pots and carriers have unique identifiers, which make it possible to treat each plant individually in transit from its position in the stationary growth area of the system to the weighing and irrigation stations and the imaging cabins.
Currently, the system includes three weighing and irrigation stations with rotating platforms and the possibility to apply water and up to three different solutions. Soil water or nutrient deficit conditions can thus be imposed on plants.
The imaging cabins are enclosed areas with camera-adapted lighting conditions and a lift with a rotating platform. At present, three camera systems are available in the cabins. The first one consists of RGB cameras in a multi-view imaging setup for the three-dimensional reconstruction of plants and the measurement of growth-related phenotypic traits. Plant physiology-related traits are measured or approximated by exploring a larger stretch of the electromagnetic spectrum. A thermal infrared camera captures energy emitted at 8-13 µm. The corresponding contextual plant and leaf temperature is used as a proxy for plant water use behavior. A state-of-the-art hyperspectral imaging system, consisting of a visible to near-infrared camera (VNIR, 400-1000 nm) and a short-wave infrared camera (SWIR, 1000-2500 nm), constitutes a novel tool for close-range sensing of plant physiological traits based on reflectance spectra captured on whole-plant and individual leaf level.
Smart features of the infrastructure include a 'handling zone', where the system can bring and retrieve a requested number of plants belonging for example to a certain genotype or treatment. As the handling zone is accessible by users of the system, it allows for visual observations of plants or manual actions on plants, such as the measurement of specific plant traits or the extraction of plant parts for molecular or biochemical analyses. A second smart feature of the system is the possibility to load external plants into the system, for example plants grown in another greenhouse compartment or growth chamber, in order to have them imaged and/or treated at the weighing and irrigation stations.
Environmental parameters, including air temperature, relative humidity and light intensity (photosynthetically active radiation), are continuously monitored in the greenhouse to direct the greenhouse heating, ventilation, humidification and lighting system, but also to support genotype-environment interaction studies in greenhouse conditions.
Experimental setups, experiment metadata and results are managed by PIPPA (PSB Interface for Plant Phenotype Analysis), the central user interface and database.
PHENOVISION has been developed and constructed in collaboration with SMO and is financially supported by a grant of the Hercules Foundation (Belgium) awarded to Professor Dirk Inzé.