Automated Renal Tubule and Cyst Quantification (ARTCyQ)

ARTCyQ is a modular image analysis framework designed for the automated detection, segmentation, and classification of renal tubules and cysts in kidney sections using immunofluorescence imaging.

The ARTCyQ pipeline combines:

• Deep learning-based segmentation (Omnipose) for tubule detection

• Pixel classification (QuPath) for cyst detection, and nuclear and cytoplasmic segmentation

• Custom Groovy scripts for automated processing

This repository contains:

• A general guideline (1-ARTCyQ_guideline.docx/.pdf) providing a detailed methodological description, including training procedures, validation strategy, workflow execution, limitations, and extensibility
• A demo project (ARTCyQ_demo/) providing a ready-to-use QuPath project structure required to run the pipeline, including one control and one cystic kidney section annotated with ARTCyQ outputs
• A directory (ARTCyQ_templates/) containing the pixel classifiers, the trained Omnipose model, and the Groovy scripts; each script is accompanied by documentation describing the workflow and user-adjustable parameters
• An Omnipose training package (Omnipose_training/) containing the trained model and annotated datasets used for both training and validation  

ARTCyQ is designed to run within a standard QuPath project. To set up a new project:

1. Create a new QuPath project
2. Copy the following folders from the Template/ directory into your project:
   • classifiers/
   • models/
   • scripts/
3. Add your images to the data/ folder
4. Ensure fluorescence channel names match expected marker names (i.e., LTL, Corin, DBA, THP, NCC)
5. Execute the scripts as described in the 1-ARTCyQ_guideline.docx/.pdf and User-tunable parameters.pdf

Important Notes:

• The folder structure and relative paths must be preserved to ensure proper script execution
• Some parameters may require adjustment depending on image acquisition batches and staining conditions; instructions for optimizing these settings are detailed in the script workflow documentation
• GPU support is recommended for optimal performance
• ARTCyQ is designed as a flexible and extensible framework that can be adapted to a wide range of applications through model retraining, integration of additional classifiers, and customization of analysis workflows

ARTCyQ validated on biop-desktop:0.1.5