DopplerQualityClassification

This repository hosts the scripts used for training and deploying the deep learning model for quality classification of 1-D Doppler Ultrasound signals. The model is detailed in the paper Motie-Shirazi M, Nikookar S, Ahmad M, Rafiei A, Sameni R, Rohloff P, Clifford GD, Katebi N. Real-time quality feedback on Doppler data for community midwives using edge-AI. Mach Learn Health. 2025 Dec.

Please cite this work when using the work in this repository

Repository Contents

1. Signal_Quality_Train.py

• This Python script includes the deep learning model training procedure for signal quality classification of Doppler recordings.

2. Helper_Functions.py

• A Python script that provides auxiliary functions essential for the training of the model.

3. Model_Application_Demo.ipynb

• A Jupyter Notebook that illustrates how to load and employ the model for classification tasks.

4. saved_model Folder

• Contains the trained deep learning model utilized for quality classification.

5. Sample_Segments Folder

• Includes a 3.75-second signal sample from each of the quality classes: 'Good', 'Poor', 'Interference', 'Talking', and 'Silent', which can be used to test the model.