FHRDiff 🚀

About 📚

This repository contains code for our paper "FHRDiff: Leveraging Diffusion Models for Conditional Fetal Heart Rate Signal Generation" presented at BIBM 2024.

Dataset 📈

The dataset used for this project can be downloaded from the following open-source repository: CTU-UHB CTG Database.

Files 📂

• classify.py: Classify the original FHR data into normal/abnormal based on the PH value.
• spectrogram: Generate the PRSA curve and its spectrogram.
• preprocess.py: Preprocess the original FHR signals, including removing gaps, interpolation, detrending, and segmenting the signals to specific lengths.
• train.py: Code for training the unconditional and conditional models.
• inference.py: Code for using the trained model to synthesize FHR signals.

Getting Started ✅

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/FHRDiff.git