Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Cardiac Output Prediction from Echocardiograms: Self-Supervised Learning with Limited Data

Adson Duarte, Davide Vitturini, Emanuele Milillo, Andrea Bragagnolo, Carlo Alberto Barbano, Riccardo Renzulli, Michele Cannito, Federico Giacobbe, Francesco Bruno, Ovidio de Filippo, Fabrizio D'Ascenzo, Marco Grangetto

TL;DR

This work proposes a self-supervised learning (SSL) pretraining strategy based on SimCLR to improve CO prediction from apical four-chamber echocardiographic videos, and shows that SSL mitigates overfitting and improves representation learning.

Abstract

Cardiac Output (CO) is a key parameter in the diagnosis and management of cardiovascular diseases. However, its accurate measurement requires right-heart catheterization, an invasive and time-consuming procedure, motivating the development of reliable non-invasive alternatives using echocardiography. In this work, we propose a self-supervised learning (SSL) pretraining strategy based on SimCLR to improve CO prediction from apical four-chamber echocardiographic videos. The pretraining is performed using the same limited dataset available for the downstream task, demonstrating the potential of SSL even under data scarcity. Our results show that SSL mitigates overfitting and improves representation learning, achieving an average Pearson correlation of 0.41 on the test set and outperforming PanEcho, a model trained on over one million echocardiographic exams. Source code is available at https://github.com/EIDOSLAB/cardiac-output.

Cardiac Output Prediction from Echocardiograms: Self-Supervised Learning with Limited Data

TL;DR

This work proposes a self-supervised learning (SSL) pretraining strategy based on SimCLR to improve CO prediction from apical four-chamber echocardiographic videos, and shows that SSL mitigates overfitting and improves representation learning.

Abstract

Cardiac Output (CO) is a key parameter in the diagnosis and management of cardiovascular diseases. However, its accurate measurement requires right-heart catheterization, an invasive and time-consuming procedure, motivating the development of reliable non-invasive alternatives using echocardiography. In this work, we propose a self-supervised learning (SSL) pretraining strategy based on SimCLR to improve CO prediction from apical four-chamber echocardiographic videos. The pretraining is performed using the same limited dataset available for the downstream task, demonstrating the potential of SSL even under data scarcity. Our results show that SSL mitigates overfitting and improves representation learning, achieving an average Pearson correlation of 0.41 on the test set and outperforming PanEcho, a model trained on over one million echocardiographic exams. Source code is available at https://github.com/EIDOSLAB/cardiac-output.
Paper Structure (9 sections, 1 equation, 2 figures, 1 table)

This paper contains 9 sections, 1 equation, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Works
  3. Methodology
  4. Pretraining
  5. Fine-tuning on the downstream task
  6. Experimental Results
  7. Dataset and preprocessing
  8. Model training
  9. Conclusion

Figures (2)

  • Figure 1: Test-set Pearson correlation versus pretraining dataset size (log scale, number of videos) for ViViT-SSL (ours), PanEcho-R, and ViViT-SL.
  • Figure 2: Predicted versus ground-truth CO on the test set for ViViT-SSL (green) and PanEcho-R (blue). Solid lines show linear regression fits; shaded regions denote 90% confidence intervals.