HIRROS: Spatio-Temporal Root System Architecture Phenotyping dataset

Spatiotemporal (2D+t) Root System Architecture Dataset of Arabidopsis SeedlingsCe dataset a été utilisé pour le papier suivant: Do Segmentation Objectives Match Reconstruction Goals? An Epoch-Wise Analysis of Root Phenotyping Pipelines This dataset contains high-resolution, time-lapse imaging data and structural reconstructions of Arabidopsis thaliana root systems. The images were acquired using the High Resolution ROot Scanner (HIRROS) automated platform. The dataset is designed to evaluate and train root system architecture reconstruction algorithms, it contains registered time serie images, time-informed masks, automatically generated graphs, and manually expertized ground-truth graphs. The data processing and automatic graph generation were performed using the RootSystemTracker pipeline. Citation If you use this dataset, please cite the following original publication describing the methodology, the RootSystemTracker pipeline, and the data acquisition process: Fernandez, R., Crabos, A., Maillard, M., Nacry, P., & Pradal, C. (2022). High-throughput and automatic structural and developmental root phenotyping on Arabidopsis seedlings. Plant Methods, 18(127). DOI: https://doi.org/10.1186/s13007-022-00960-5 For the RSML (Root System Markup Language) data format: Lobet, G., Pound, M. P., Diener, J., Pradal, C., Draye, X., Godin, C., Javaux, M., Leitner, D., Meunier, F., Nacry, P., Pridmore, T. P. & Schnepf, A. (2015). Root system markup language: toward a unified root architecture description language. Plant Physiology, 167(3), 617-627. DOI : https://doi.org/10.1104/pp.114.253625 Dataset Structure The dataset is partitioned into three main directories designed for machine learning and algorithmic evaluation workflows: Train/: Contains time-series data and corresponding root system models intended for model training. Val/: Contains validation data for hyperparameter tuning and model evaluation. Test/: Contains unseen data for model evaluation, not all plants are segmented and / or reconstructed. Each directory contains subfolders corresponding to individual time series boxes (e.g., 230629PN008). File Descriptions Within each sample directory, you will find the following files: 22_registered_stack.tif : A cropped, time-lapse image stack of the root system. These images have undergone rigid and dense non-linear registration to correct misalignments and ensure root immobility across the time series. 31_mask_at_t1.tif : A binary mask used as the base input for the RSA reconstruction step in the RootSystemTracker pipeline. 40_date_map.tif : A time-indexed image mask acting as the ground truth for segmentation. Each pixel value corresponds to its temporal emergence index. 61_before_expertized_graph.rsml : The raw, uncorrected output generated automatically by the RootSystemTracker pipeline from the corresponding masks. It is formatted as RSMLs. 61_graph.rsml : The ground-truth RSML file. This represents the precise geometry and topology of the root system. It was obtained by taking 61_before_expertized_graph.rsml and manually correcting it by an expert at LEPSE, INRAE Montpellier, using a dedicated 2D+t model edition Fiji plugin. InfoSerieRootSystemTracker.csv : A metadata file containing parameters and information required for the reconstruction process via the RootSystemTracker pipeline.