InterstitialFibrosis

Implementation of the methods described in "Image Analysis for Non-Neoplastic Kidney Disease: Utilizing Morphological Segmentation to Improve Quantification of Interstitial Fibrosis". Compute interstitial fibrosis from Sirius-red-stained whole-slide images (WSI) using QuPath-style GeoJSON annotations for cortex, tubules, and glomeruli. The tool generates fibrosis masks, summary reports, and (optionally) per-tile debug outputs.

Expected folder structure

ProjectRoot/
├── Images/
│   └── Image1.svs
├── Annotations/
│   ├── Cortex/
│   │   └── Image1.geojson
│   ├── Glomeruli/
│   │   └── Image1.geojson
│   └── Tubules/
│       └── Image1.geojson
├── FibrosisReport.py
└── Utils/
    ├── __init__.py
    ├── geojson_mask.py
    ├── zarr_helpers.py
    ├── mask_reader.py
    ├── image_utils.py
    ├── fibrosis_methods.py
    ├── paths.py
    ├── tile_selection.py
    └── disk_cache.py

Usage

Environment

conda env create -f Env/FibrosisReport.yml

CLI

python FibrosisReport.py --image_path [Required] -f/-d -RAM/-DISK --method [Required] --report [Optional] --no-mask [Optional] --post_processing [Optional]

Options

• --image_path (required)
  Path to a single image file (-f) or a directory (-d).
• -f / -d (required)
  Treat image_path as file or directory.
• -RAM / -DISK (optional)
  -RAM (default): store only full mask in RAM.
  -DISK: additionally saves per-tile debug outputs to disk.
• --method (required)
  stain_decon
red_green_filtering
• --report (optional)
  simple (default): per-image summary CSV.
  full: adds tile-level CSV and stores tile output paths.
• --no-mask (optional)
  Disable saving the final fibrosis mask TIFF.
• --post_processing (optional)
  True (default): remove small components.
  False: keep raw output.

---

Examples

1) Run on a single file (RAM mode)

python FibrosisReport.py --image_path "ProjectRoot/Images/Image1.svs" -f -RAM --method stain_decon --report simple --post_processing True

2) Run on all images in a folder (full report)

python FibrosisReport.py --image_path "ProjectRoot" -d -RAM --method red_green_filtering --report full --post_processing True

3) DISK mode (save per-tile bundles)

python FibrosisReport.py --image_path "ProjectRoot/Images/Image1.svs" -f -DISK --method stain_decon --report full --post_processing True

---

License

Shield:

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.