Investigating Shortcut Learning in Medical Image Analysis: Insights from the ShorT Framework

by Katarina Kraljevic katrk@itu.dk Paula Menshikoff pmen@itu.dk

Overview

This repository contains the code and data for the BSc project "Shortcuts Problem in Medical Image Analysis." The project explores the issue of shortcuts in medical image analysis, specifically focusing on chest X-ray images.

Repository Structure

• EDA/: Exploratory Data Analysis scripts and notebooks.
• Meetings/: Notes and materials from project meetings.
• shortcuts-chest-xray/: Main codebase for the project.

Requirements

• Python 3.x
• Jupyter Notebook
• Required Python libraries: listed in requirements.txt

Installation

Clone the repository:

git clone https://github.com/katarinakoke/Shortcuts-Problem-in-Medical-Image-Analysis.git
cd Shortcuts-Problem-in-Medical-Image-Analysis