WSI Viewer with Segmentation Visualization

This project extends a whole slide image (WSI) viewer to visualize segmentation results, built with Electron, Next.js, TypeScript, and FastAPI.

Video Demo

Features

• WSI viewing with support for zooming and panning
• Segmentation results overlay on the WSI
• Management sidebar for reviewing all segmentation results
• User annotation functionality (cropping regions and changing annotation colors)
• Modern UI with Tailwind CSS and React components

Setup Instructions

Prerequisites

• Node.js (v14+)
• Python (v3.8+)
• Git

Installation

1. Clone the repository:

   git clone <repository-url>
   cd wsi-viewer

2. Install frontend dependencies:

   cd wsi-viewer-app
   npm install

3. Install backend dependencies:

   cd ../wsi-backend
   pip install -r requirements.txt

4. Download test data (if needed):

   wget https://openslide.cs.cmu.edu/download/openslide-testdata/Aperio/CMU-1.svs

Running the Application

1. Start the backend server:

   cd wsi-backend
   python main.py

2. Start the frontend application:

   cd wsi-viewer-app
   npm run dev

3. The application should open automatically in Electron. If not, open it manually according to instructions in the terminal.

Project Structure

• wsi-backend/: FastAPI backend for handling WSI data and segmentation
• wsi-viewer-app/: Electron + Next.js frontend application
• sample_code/: Reference implementation

Implementation Details

This project addresses the requirements of visualizing segmentation results on whole slide images by:

1. Using Electron and Next.js for a responsive desktop application
2. Implementing an overlay system that shows:
   • Segmentation centroids as dots at low zoom levels
   • Detailed contours when zoomed in past a threshold
3. Providing a management interface for reviewing all segmentation results
4. Supporting user annotations within cropped regions

Technologies Used

• Frontend: Electron, Next.js, TypeScript, OpenSeadragon, Tailwind CSS
• Backend: FastAPI, Python
• Data Handling: HDF5 for segmentation results

Future Improvements

• Performance optimizations for large WSIs
• Additional annotation tools
• Support for multiple segmentation masks
• Collaborative viewing and annotation