MISO: Medical Image Segmentation for Organoids

This project is a full-stack application designed for the segmentation of medical images, with a specific focus on organoids. It utilizes a deep learning model on the backend and provides a user-friendly interface on the frontend for image upload and analysis.

Problem Statement

Medical image analysis is a critical component of diagnostics and research. The accurate segmentation of specific structures, such as organoids, from complex images is a time-consuming and challenging task when done manually. This project aims to automate this process by using a deep learning model (Attention U-Net) to perform precise and efficient image segmentation, saving time and improving accuracy for researchers. This is what I came up with when I saw the dataset.

Project Architecture

The project is divided into two main components:

• Backend: A Python-based API built with FastAPI. It handles image processing, runs the segmentation model, and serves the results. The deep learning model is an Attention U-Net implemented using PyTorch.
• Frontend: A modern, responsive user interface built with React and Vite. It allows users to upload images and view the segmentation results returned by the backend.

How to Run the Project

To get the application running locally, you will need to start both the backend server and the frontend development server.

1. Backend (FastAPI)

The backend is managed using uv, a fast Python package installer and resolver.

Prerequisites:

• Python 3
• uv installed (pip install uv)

Instructions:

1. Navigate to the project root directory:

   cd path/to/miso

2. Create a virtual environment and install dependencies:

   # Old school
   uv venv
   uv pip install -r requirements.txt
   
   # One line
   uv sync

   (Note: If a requirements.txt is not present, you can generate one from pyproject.toml or install dependencies directly with uv pip install -e .)

3. Start the server: The application is run using uvicorn. From the root directory, run:

   uvicorn main:app --host 0.0.0.0 --port 8000 --reload

   The API will be available at http://localhost:8000. You can go to https://localhost:8000/docs to see list of endpoints and how to use them.

2. Frontend (React)

The frontend is a standard React application managed with npm.

Prerequisites:

• Node.js

Instructions:

1. Navigate to the frontend directory:

   cd path/to/miso/frontend

2. Install dependencies:

   npm install

3. Start the development server:

   npm run dev

   The frontend application will be available at http://localhost:5173 (or another port if 5173 is busy). You can now open this URL in your browser to use the application.