🩺🔬 SkinScan

SkinScan is a skin cancer detector web application that allows users to upload images of their skin conditions. Our system uses a multi-class ML-model that analyses uploaded images, in conjunction with other features such as age and sex, to determine the probable type of condition and whether it is benign (noncancerous) or malignant (cancerous).

For legal and ethical concerns, the aim of this system is not to provide any medical diagnosis but rather a recommendation on whether the user should seek out professional medical assistance. The goal is to minimise false negatives (i.e., optimise recall) while retaining acceptable overall model accuracy.🎯 This ensures users can trust the model’s predictions for benign conditions. The model slightly overrepresents malignant conditions to avoid missing true positives, as the risks of missing malignant conditions far outweigh the inconvenience of recommending medical advice for benign conditions.

To enhance transparency 🔍 and build user trust 🤝, the results include AI explainability measures.🤖💡 These measures display a percentage score for each feature’s relative impact on the prediction, along with a heatmap overlay highlighting the areas of the input image that our ML model focused on during processing.

The system’s web application includes an admin panel UI for administrator users. This panel provides access to system analytics📊 and other functionalities, such as managing the ML-pipeline to train new models or replace the current active model used for running inference on user data. Administrators can view previous model versions, review their hyperparameters, and compare performance metrics across different versions using visual graphs. The admin panel also provides detailed insights into the usage and accuracy of the system, helping developers and healthcare professionals make informed improvements. This ensures the tool remains accurate, effective, and trustworthy.

📑 Table of Contents

• Svelte Web-app [Frontend]
• Running Django [Backend]
  • Create .env File in Repository Root Folder
  • Run development server
    • macOS/Linux:
    • Windows WSL:
  • Database Migrations
  • Unit Tests
  • Deactivating the Virtual Environment
• Installation [Backend]
  • macOS/Linux:
    • Set Up Python Virtual Environment & Install Dependencies
  • Windows WSL:
    • Installing Python 3.11 on Ubuntu WSL
    • Set Up Python Virtual Environment & Install Dependencies
  • Optional: Enable Nvidia CUDA GPU Support [Linux / WSL]
    • Step 1: Update Nvidia Drivers
    • Step 2: Install CUDA Toolkit
    • Step 3: Install cuDNN
    • Step 4: Run Test Script to Verify GPU Utilization
• Development team

Svelte Web-app [Frontend]

To run locally, refer to the instructions inside the Client directory README.

Running Django [Backend]

Create .env File in Repository Root Folder

Make sure the .env file contains the following:

# Django secret key
SECRET_KEY = <KEY_VALUE>

# Use "False" for production
DEBUG = "True"

Run development server

macOS/Linux:

1. Navigate to the Django project root folder

   cd server
   

2. Run the Django development server

   python3 manage.py runserver
   

3. Open browser and navigate to:

   http://127.0.0.1:8000
   

Windows WSL:

1. Navigate to the Django project root folder

   cd server
   

2. Run the Django development server

   python manage.py runserver
   

3. Open browser and navigate to:

   http://127.0.0.1:8000
   

Database Migrations

If any changes are made to the Django models (database schemas), the changes need to be migrated to the database(s). Execute the following commands from the Django project root folder:

python3 manage.py makemigrations
python3 manage.py migrate
python3 manage.py migrate --database=db_images

Unit Tests

To run the Django unit tests, execute the following commands from the Django project root folder:

python3 manage.py test    

Deactivating the Virtual Environment

Once you are done, deactivate the Python virtual environment using:

deactivate

Installation [Backend]

macOS/Linux:

Set Up Python Virtual Environment & Install Dependencies

1. Navigate to the repository root folder in your terminal

   cd /path/to/repository
   

2. Create a Python virtual environment

   python3 -m venv venv
   

3. Activate the virtual environment

   source venv/bin/activate
   

4. Install the required dependencies

   on Linux:

   pip install -r requirements.txt
   

   on macOS:

   pip install -r requirements-mac.txt
   

Windows WSL:

Note: this guide is written for WSL2 using Ubuntu 22.04 LTS (Jammy).

Installing Python 3.11 on Ubuntu WSL

Set Up Python Virtual Environment & Install Dependencies

1. Navigate to the repository root folder in your terminal

   cd /path/to/repository
   

2. Create a Python virtual environment

   python3.11 -m venv venv
   

3. Activate the virtual environment

   source venv/bin/activate
   

4. Upgrade pip inside the virtual environment

   pip install --upgrade pip
   

5. Install the required dependencies

   pip install -r requirements.txt
   

Optional: Enable Nvidia CUDA GPU Support [Linux / WSL]

In order to utilize the GPU for TensorFlow operations, additional setup is needed.

wget https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-wsl-ubuntu.pin

sudo mv cuda-wsl-ubuntu.pin /etc/apt/preferences.d/cuda-repository-pin-600

wget https://developer.download.nvidia.com/compute/cuda/12.3.2/local_installers/cuda-repo-wsl-ubuntu-12-3-local_12.3.2-1_amd64.deb

sudo dpkg -i cuda-repo-wsl-ubuntu-12-3-local_12.3.2-1_amd64.deb

sudo cp /var/cuda-repo-wsl-ubuntu-12-3-local/cuda-*-keyring.gpg /usr/share/keyrings/

sudo apt-get update

sudo apt-get -y install cuda-toolkit-12-3

nvcc --version

Development Team👩‍💻👨‍💻

The project has been developed over the course of 8 weeks by the following:

• Kaisa Arumeel
• Amirpooya Asadollahnejad
• Erik Lindstrand
• Arvin Rahimi
• Konstantinos Rokanas
• Alexander Säfström