Artery Feature Extraction Pipeline

This repository provides a fully automated pipeline to measure arterial wall thickness (intima & media) and compute derived morphometric features from histology whole-slide images (WSIs) and their corresponding GeoJSON annotations.

It reads artery‐specific annotations (media, intima, lumen contours), computes per-artery thickness maps, area fractions, and intima-to-media ratios, and saves per-artery feature summaries in CSV and JSON formats.

Overview

Component	Description
Input WSIs	.svs or .ndpi whole-slide images (e.g., trichrome stain)
Input annotations	.geojson files containing labeled polygons (“Media”, “Intima”, “Lumen”)
Outputs	JSON file (thickness.json) with raw measurements, and CSV file (artery_features_no_labels.csv) with aggregated morphometric features
Visualization	Overlay figures saved under results/figures/

Directory Structure

artery_thickness/
├── artery_analysis.py           # Main CLI script (this file)
├── helper.py                    # Supporting geometry & analysis utilities
├── environment.yml             # Optional dependency list
├── data/
│   ├── wsi/                     # Input WSIs (.svs/.ndpi)
│   └── ann_geojson/             # Corresponding annotation files (.geojson)
└── results/
    ├── thickness.json           # Raw per-artery measurements
    ├── artery_features_no_labels.csv  # Aggregated feature summary
    └── figures/                 # Visualization outputs

Usage

You can run the pipeline with a single command using the CLI interface.

Example: python artery_analysis.py
--wsi_dir /path/to/wsi
--ann_dir /path/to/annotations
--save_dir /path/to/save/results

Output Files

File	Description
thickness.json	Raw artery-level thickness and area measurements (per artery per WSI).
artery_features_no_labels.csv	Final computed morphometric feature table.
figures/	Optional visualization overlays for QA.

Each row in the CSV corresponds to a unique (WSI_ID, Artery_ID) pair, including features like:

Feature	Description
Thickness_Media_um	Media layer thickness (μm)
Thickness_Intima_um	Intima layer thickness (μm)
Media Area (microm2)	Media area (μm²)
Intima Area (microm2)	Intima area (μm²)
Lumen Area (microm2)	Lumen area (μm²)
Ratio Intima/Media Area	Ratio of intima to media area
Media Area Frac	Fraction of total artery area occupied by media
Intima Area Frac	Fraction of total artery area occupied by intima
Lumen Area Frac	Fraction of total artery area occupied by lumen

Computation Flow

The artery wall thickness analysis pipeline proceeds through the following stages:

1. Load Annotations

   • Reads polygonal regions from .geojson annotation files.
   • Cleans coordinate data and filters polygons based on their labels (Media, Intima, Lumen).

2. Determine Region of Interest (ROI)

   • Computes the global bounding box encompassing all annotated artery polygons.
   • Defines the region of the slide to extract for downstream analysis.

3. Read Slide Region

   • Extracts the corresponding tissue region from the whole-slide image (WSI) using OpenSlide.
   • Converts RGBA images to RGB for consistency in visualization and processing.

4. Polygon Processing

   • Transforms global annotation coordinates to local (crop-level) coordinates.
   • Crops and overlays contours on the extracted image for visual validation.

5. Measurement

   • Computes pixel-level areas for lumen, intima, and media.
   • Converts pixel areas to physical units (µm²) using the slide’s microns-per-pixel (MPP) metadata.
   • Measures intima and media thickness along evenly spaced radial angles.

6. Feature Extraction

   • Aggregates morphometric descriptors such as mean, standard deviation, and ratios.
   • Computes structural metrics like area fractions and intima/media ratios.
   • Outputs results in both structured CSV and JSON formats for easy downstream analysis.

---

Author

Harishwar Reddy Kasireddy
Ph.D. Student, Electrical & Computer Engineering
University of Florida — Sarder Lab (Computational Pathology)
[email protected]

Relation to Prior Work

This repository is a re-implementation (engineering reproduction) of the core measurement pipeline described in:

Zhou J., Li X., Demeke D., Dinh T.A., et al. “Characterization of arteriosclerosis based on computer-aided measurements of intra-arterial intima-media thickness.” Journal of Medical Imaging 11(5):057501, 2024. doi:10.1117/1.JMI.11.5.057501. ([ResearchGate][1])

and a modularized version of this reposiotry https://github.com/code-by-jin/Measurements-of-Intra-Arterial-Thickness.git