Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Automatic Cardiac Pathology Recognition in Echocardiography Images Using Higher Order Dynamic Mode Decomposition and a Vision Transformer for Small Datasets

Andrés Bell-Navas, Nourelhouda Groun, María Villalba-Orero, Enrique Lara-Pezzi, Jesús Garicano-Mena, Soledad Le Clainche

TL;DR

This work tackles automatic cardiac pathology recognition from echocardiography sequences, addressing data scarcity by combining Higher Order Dynamic Mode Decomposition (HODMD) based data augmentation with a Vision Transformer (ViT) trained from scratch on a large, homogenized database. The system consists of a data-generation stage (ROI-focused homogenization and modal decomposition-based augmentation) and a pathology-prediction stage (SPT-augmented ViT with Locality Self Attention, fused predictions across sequences). Empirical results on a mouse-model echo dataset show that HODMD-derived features and ViT-based learning yield superior accuracy, including up to about 0.77 per-sequence accuracy, outperforming pretrained CNN baselines. The approach demonstrates real-time feasibility and suggests scalability with larger datasets could further improve robustness and performance.

Abstract

Heart diseases are the main international cause of human defunction. According to the WHO, nearly 18 million people decease each year because of heart diseases. Also considering the increase of medical data, much pressure is put on the health industry to develop systems for early and accurate heart disease recognition. In this work, an automatic cardiac pathology recognition system based on a novel deep learning framework is proposed, which analyses in real-time echocardiography video sequences. The system works in two stages. The first one transforms the data included in a database of echocardiography sequences into a machine-learning-compatible collection of annotated images which can be used in the training stage of any kind of machine learning-based framework, and more specifically with deep learning. This includes the use of the Higher Order Dynamic Mode Decomposition (HODMD) algorithm, for the first time to the authors' knowledge, for both data augmentation and feature extraction in the medical field. The second stage is focused on building and training a Vision Transformer (ViT), barely explored in the related literature. The ViT is adapted for an effective training from scratch, even with small datasets. The designed neural network analyses images from an echocardiography sequence to predict the heart state. The results obtained show the superiority of the proposed system and the efficacy of the HODMD algorithm, even outperforming pretrained Convolutional Neural Networks (CNNs), which are so far the method of choice in the literature.

Automatic Cardiac Pathology Recognition in Echocardiography Images Using Higher Order Dynamic Mode Decomposition and a Vision Transformer for Small Datasets

TL;DR

This work tackles automatic cardiac pathology recognition from echocardiography sequences, addressing data scarcity by combining Higher Order Dynamic Mode Decomposition (HODMD) based data augmentation with a Vision Transformer (ViT) trained from scratch on a large, homogenized database. The system consists of a data-generation stage (ROI-focused homogenization and modal decomposition-based augmentation) and a pathology-prediction stage (SPT-augmented ViT with Locality Self Attention, fused predictions across sequences). Empirical results on a mouse-model echo dataset show that HODMD-derived features and ViT-based learning yield superior accuracy, including up to about 0.77 per-sequence accuracy, outperforming pretrained CNN baselines. The approach demonstrates real-time feasibility and suggests scalability with larger datasets could further improve robustness and performance.

Abstract

Heart diseases are the main international cause of human defunction. According to the WHO, nearly 18 million people decease each year because of heart diseases. Also considering the increase of medical data, much pressure is put on the health industry to develop systems for early and accurate heart disease recognition. In this work, an automatic cardiac pathology recognition system based on a novel deep learning framework is proposed, which analyses in real-time echocardiography video sequences. The system works in two stages. The first one transforms the data included in a database of echocardiography sequences into a machine-learning-compatible collection of annotated images which can be used in the training stage of any kind of machine learning-based framework, and more specifically with deep learning. This includes the use of the Higher Order Dynamic Mode Decomposition (HODMD) algorithm, for the first time to the authors' knowledge, for both data augmentation and feature extraction in the medical field. The second stage is focused on building and training a Vision Transformer (ViT), barely explored in the related literature. The ViT is adapted for an effective training from scratch, even with small datasets. The designed neural network analyses images from an echocardiography sequence to predict the heart state. The results obtained show the superiority of the proposed system and the efficacy of the HODMD algorithm, even outperforming pretrained Convolutional Neural Networks (CNNs), which are so far the method of choice in the literature.
Paper Structure (10 sections, 2 equations, 7 figures, 5 tables)

This paper contains 10 sections, 2 equations, 7 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Cardiac pathology recognition system
  3. Medical Database Creation
  4. Higher Order Dynamic Mode Decomposition
  5. Pathology Recognition
  6. Database
  7. Results
  8. Analysis of types of data
  9. Comparison With Alternative Algorithms
  10. Conclusions

Figures (7)

  • Figure 1: Block diagram of the proposed cardiac pathology recognition system, with representations of the echocardiography images with non-heart regions (the electrocardiogram, medical information, and the black background).
  • Figure 2: Block diagram of the Medical Database Creation stage, composed of two phases: Data Homogenization, and Modal Decomposition-based Data Generation. The former phase is further divided into (a) Spatial homogenization and (b) Temporal homogenization.
  • Figure 3: Block diagram of the HODMD algorithm applied on a sequence of echocardiography images.
  • Figure 4: Block diagram of the Pathology Recognition stage.
  • Figure 5: Architecture of the proposed deep neural network. (a) The input: an image from an echocardiography video sequence. (b) The SPT module for more embedding of spatial information in each visual token. (c) The ViT with multi-head attention based on LSA for more locality inductive bias. (d) The predicted heart state of the input image.
  • ...and 2 more figures