Published April 17, 2026 | Version v1
Dataset Open

Phase-contrast image dataset for morphological evaluation of SH-SY5Y differentiation after rosmarinic and retinoic acid treatments

Authors/Creators

  • 1. Department of Pharmacology, Pharmacotherapy and Toxicology, Faculty of Pharmacy, Medical University – Sofia, Bulgaria
  • 2. Research Institute of Innovative Medical Science, Medical University – Sofia, Bulgaria

Description

This dataset contains raw phase-contrast microscopy images, image-level metadata, an ilastik pixel-classification project/model, ilastik training images, preprocessed ilastik training images, example expected ilastik outputs, and compact representative analysis outputs generated during morphological analysis of SH-SY5Y cells treated with rosmarinic acid and retinoic acid at different treatment concentrations, FBS concentrations, cell seeding densities and treatment times.

The dataset supports the expected publication entitled "Morphological evaluation of the differentiation status of SH-SY5Y cells upon rosmarinic and retinoic acid treatments". The associated image-analysis workflow is designed to quantify neurite-associated morphology from phase-contrast images, including segmentation-derived neurite, cell-body, and background regions, ROI-level neurite features, skeleton-derived features, image-level summaries, and downstream statistical/visual interpretation.

Technical info (English)

Images were taken with an 8 megapixel CCD (Charged Coupled Device) digital camera
(Optikam Pro 8LT - 4083.18LT), mounted on an inverted Optika XDS-2 microscope, illuminated by a X-LED8TM system. We applied the following microscopy settings: phase contrast mode and 100x magnification. In order to diminish cell density variation due to intrawell cell distribution gradients, all images were taken after centering the microscopic field equidistantly from the center and walls of each well. Before taking images, the focus of microscopy was always brought close to the periphery of cells so that contrast at cell or monolayer borders is optimal and reproducible among images.

The dataset has been subjected to analysis with the companion software workflow integrating R, Fiji/ImageJ, and ilastik, which is available in the GitHub repository:

https://github.com/YordanovLab/PhaseContrastNeuriteAnalyzer

The software provides a browser-based interface for configuring the analysis, preparing images, applying Fiji preprocessing, running ilastik segmentation, validating segmentation labels, optimizing neurite cutoff models, applying selected cutoff models to larger image sets, and generating downstream visualization and statistical analysis outputs.

The included ilastik project/model is intended for the same type of Fiji-preprocessed phase-contrast images used in this dataset. In the workflow, raw images are first standardized by Fiji/ImageJ preprocessing, including generation of red/green averaged intensity images. The ilastik pixel classifier is then trained/applied using three classes: background, cell bodies, and neurites. The output masks are subsequently interpreted by the analysis software using the confirmed label mapping.

To recreate the analysis after downloading this dataset, clone the GitHub repository and copy the archive contents into the repository as follows: copy `raw_images/*` into `pipeline_inputs/`; copy `metadata/IN_sample_metadata_zenodo_paths.csv` to `pipeline_inputs/IN_sample_metadata.csv`; copy `ilastik/model/*.ilp` into `models/`; copy ilastik training/example folders into `example_data/ilastik_training/`; and copy `example_outputs/` into `example_data/example_outputs/`. Then copy `config/pipeline_settings.example.env` to `config/pipeline_settings.env` and point `INPUT_MASTER_DIR`, `SAMPLE_METADATA_FILE`, and `ILASTIK_PROJECT_FILE` to those restored files.

Other (English)

Keywords: SH-SY5Y; neurite analysis; phase-contrast microscopy; rosmarinic acid; retinoic acid; neuronal differentiation; cell morphology; image analysis; Fiji; ImageJ; ilastik; R; segmentation; skeletonization; pixel classification.

Funding: This study was funded by the European Union - NextGenerationEU through the National Recovery and Resilience Plan of the Republic of Bulgaria, project No. BG-RRP-2.004-0004-C01.

Notes

Troubleshooting note for Linux/WSL users: after unzipping the archive, some folders may lack the directory execute/search permission bit. Such folders can appear in `ls` but cannot be entered, causing recursive image discovery to find only part of the dataset. In the associated PhaseContrastNeuriteAnalyzer app, open Configuration and click “Check and Repair Input Folder Permissions.” Terminal fallback from the repository root: `bash ./launchers/repair_input_permissions.sh ./pipeline_inputs/raw_images`. Manual equivalent: `chmod -R u+rwX ./pipeline_inputs/raw_images`. The capital `X` adds traversal permission to directories without making all image files executable.

Files

PhaseContrastNeuriteAnalyzer_dataset_v1.0.0.zip

Files (5.8 GB)

Additional details

Software

Repository URL
https://github.com/YordanovLab/PhaseContrastNeuriteAnalyzer
Programming language
R , ImageJ Macro , Python

Biodiversity

Species
Human

References