LemnaTec Software for Phenotyping and Seed Testing
Translating images into application-specific information
The core of all LemnaTec phenotyping and seed testing systems is a comprehensive software package that enables operating the sensing equipment, storing the data and metadata, access to all records, and analysis of data. Hardware- and sensor- controls depend on the system and are specifically adapted for each product and also for customized solutions.
For all camera recordings, dedicated data processing is available. This ranges from systematic file administration over image processing to advanced machine learning algorithms. The image data processing and analysis is the key component of the phenotyping and seed testing procedures. Advanced image processing algorithms combined with machine learning tools ensure target-oriented analysis of the recorded image data. The analytical software translates digital image data into biologically relevant information.
Processing tools deliver sample-related data (examples):
|Sensor||Recorded parameters||Information to derive from parameters|
|Visible light camera||Reflectance in visible light spectrum|
|Counts, dimensions, texture, colour; growth and developmental features, stress responses|
(with corresponding excitation light and filter)
|Intensity and distribution of fluorescence light||Presence and distributions of fluorescencent pigments; stress and senescence|
|Camera for chlorophyll fluorescence dynamics (Kautsky/PAM)||Photosystem II-related parameters||Photosynthetic capacity and activity; stress, pathogen responses|
|NIR camera with 1450 nm filter||Water-content related NIR signal||Tissue moisture; water stress responses|
|IR camera||Surface heat emission||Plant temperatures, transpiration|
|Hyper-/multispectral camera||Spectrally resolved reflectance||Physiological parameters, vegetation indices|
|Laser scanner||Point clouds||3D surface, height map, inclination map, convex hull; growth and developmental features|
The parameters extracted from the camera recordings deliver phenotypic information. For this purpose, we deliver a toolbox phenotyping-related functions, e.g.:
Size, count, morphological parameters, and color of plants
Height and width
Plant organ specific parameters
Time course analyses, e.g. for growth rates
3D data of plants
Stress and pathogen responses, e.g. color changes, fluorescence changes
NIR reflectance as indicator for water content
Fluorescence intensity, e.g. biomarkers or chlorophyll
Chlorophyll fluorescence parameters as indicator of photosystem status and activity
Broad range of vegetation indices together with index calculator for customized indices – more than 60 pre-set vegetation indices available in the software package.
LemnaGrid offers access to LemnaTec customer specific machine learning based solutions.
LemnaTec software at work in plant- and seed- applications
Phenotyping of seedlings on petri dish
Assessing shoot and root traits of agar-grown seedlings with PhenoAIxpert: shoots and roots are recognized separately and measured for their individual size. In the current example, primary root length, secondary root length, and shoot area were measured for each seedling.
Artificial intelligence for seedling classification
Emerging seedlings consist of various parts, e.g. roots, shoots, or root hairs. With machine learning algorithms, SeedAIxpert can be trained to recognize each of these parts separately.
Arabidopsis growth assay
Trays with growing Arabidopsis plants were imaged with PhenoAIxpert and images were processed for plant growth, morphology and color. Individual plants were identified and phenotypic properties of their visible plant area were assessed. Such tests serve for e.g. candidate screening, genetic studies, treatment effects, environmental responses
Embryo and Endosperm are visible on the surface of maize seeds. Imaged and classified by SeedAIxpert, fractions of embryo and endosperm are measured for each seed.
Seedling emergence test
Oilseed rape germination was assessed in a seedling emergence test with seeds placed in substrate. Emerging cotyledons were recognized and measured for germination frequency and seedling quality.
In automatic phenotyping systems, users can schedule operation of the system and trigger measurements with the mounted sensors. Any automated hardware can operate in a time-scheduled mode for repeated programmed actions such as moving the sensing equipment to defined places and start camera recording. Thus, time- and space- resolved experiment planning is enabled.
All settings and recordings are stored in a central database. These data can be accessed via a dedicated interface that allows data retrieving. Retrieved data can be processed with the provided data analysis tools. The output of the analyses is corresponding to the capabilities of the cameras used for the recording. The analysis tools provide a large set of drag-and-drop functions for image processing that enable users to establish customized analysis procedures. In addition, case-specific analyses can be ordered from LemnaTec that allow one-click analyses of standard procedures, such as germination testing.
The analysis tools include machine learning and thereby give access to advanced classification even in complex cases. All processing results are stored in the database, together with the original data and metadata.
Illustration of software modules for data acquisition, processing, management and analysis operated on dedicated computer hardware
The LemnaTec software consists of the following modules
that serve for the tasks as described above:
Hardware Control and data acquisition:
The LemnaControl software module controls and monitors all required hardware components such as cameras, illumination, conveyor belts, pumps and scales. It enables trained operators to efficiently run the Scanalyzer systems with only few technical and IT background.
LemnaControl facilitates rapid and easy initiation of experiments with specific configurations on the LemnaTec systems. During an experiment, all data for an individual tray are stored in the central database.
LemnaControl facilitates flexible operating modes, as part of the overall experimental. Those can be scheduled for an entire experiment, with repeating tasks e.g. individual tasks such randomizing the tray position in the system.
Data storage and backup:
The core database structure is based on SQL Database technology (either Oracle or Postgres) and can handle very large datasets. The open interface allows direct integration into any customer IT infrastructure such as e.g. LIMS environments.
Image viewer and data management:
The LemnaBase software module facilitates rapid and easy visualization of images acquired during an experimental run. Flexible viewing schemes allow direct comparison of multiple plant simultaneously. LemnaBase further allows to import images into the database and likewise export them. New databases can be created, and data be transferred.
Image Analysis and feature extraction:
The LemnaGrid software module uses a graphical programming environment (similar to LabView or Microsoft Robotics Studio), which allows easy integration of different image analysis algorithms. The image processing pipeline extracts desired properties/features from the original image and stores results in the central database. LemnaGrid uses state-of the art machine and deep learning tools, besides classical image analysis to detect e.g. germination features.
Data Analysis and visualisation:
The LemnaExplorer software allows visualization of the all experiment-related data in the data base. Dedicated filtering allows fast sub setting of data and outlier detection. The organization into experiment-specific dashboards allows custom-made data aggregation, processing and visualization.
The LemnaExplorer software is a web-based tool (i.e. it runs in the browser) and can hence be widely accessed within the customers network, allowing flexible exploration of the data set.
LEMNATEC-OS is a complete open software system for phenotyping plants, seeds and small organisms. A suite of integrated modules enables scientists to automate phenotyping processes and analyse the resulting images. Using drag-and-drop functionality, complex algorithms can be quickly constructed, modified and tested.
The APPLICATIONS section of this website shows some of the ways this powerful software suite can be used to measure specific characteristics and traits in plant phenotyping.