Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Novel real-time arrhythmia detection model using YOLOv8

Guang Jun Nicholas Ang, Aritejh Kr Goil, Henryk Chan, Jieyi Jeric Lew, Xin Chun Lee, Raihan Bin Ahmad Mustaffa, Timotius Jason, Ze Ting Woon, Bingquan Shen

TL;DR

This study reframes arrhythmia detection as a real-time 2D object-detection problem by converting 1D ECG data into 2D canvases and applying a loss-modified YOLOv8n trained on the MIT-BIH Arrhythmia Database. The authors introduce dynamic inverse class frequency and Wise IoU v3 to address data imbalance and low-quality samples, achieving a reported mAP50 of about 0.992 and classification accuracy near 99.5% with an end-to-end inference time of ~0.002 s on a Tesla V100. They also demonstrate end-to-end real-time deployment using a low-cost ECG hardware setup and discuss practical deployment considerations and limitations, including explainability and edge performance. Overall, the work advances real-time, image-based ECG analysis and lays groundwork for edge-enabled, explainable arrhythmia detection in home and clinical settings.

Abstract

In a landscape characterized by heightened connectivity and mobility, coupled with a surge in cardiovascular ailments, the imperative to curtail healthcare expenses through remote monitoring of cardiovascular health has become more pronounced. The accurate detection and classification of cardiac arrhythmias are pivotal for diagnosing individuals with heart irregularities. This study underscores the feasibility of employing electrocardiograms (ECG) measurements in the home environment for real-time arrhythmia detection. Presenting a fresh application for arrhythmia detection, this paper leverages the cutting-edge You-Only-Look-Once (YOLO)v8 algorithm to categorize single-lead ECG signals. We introduce a novel loss-modified YOLOv8 model, fine-tuned on the MIT-BIH arrhythmia dataset, enabling real-time continuous monitoring. The obtained results substantiate the efficacy of our approach, with the model attaining an average accuracy of 99.5% and 0.992 mAP@50, and a rapid detection time of 0.002 seconds on an NVIDIA Tesla V100. Our investigation exemplifies the potential of real-time arrhythmia detection, enabling users to visually interpret the model output within the comfort of their homes. Furthermore, this study lays the groundwork for an extension into a real-time explainable AI (XAI) model capable of deployment in the healthcare sector, thereby significantly advancing the realm of healthcare solutions.

A Novel real-time arrhythmia detection model using YOLOv8

TL;DR

This study reframes arrhythmia detection as a real-time 2D object-detection problem by converting 1D ECG data into 2D canvases and applying a loss-modified YOLOv8n trained on the MIT-BIH Arrhythmia Database. The authors introduce dynamic inverse class frequency and Wise IoU v3 to address data imbalance and low-quality samples, achieving a reported mAP50 of about 0.992 and classification accuracy near 99.5% with an end-to-end inference time of ~0.002 s on a Tesla V100. They also demonstrate end-to-end real-time deployment using a low-cost ECG hardware setup and discuss practical deployment considerations and limitations, including explainability and edge performance. Overall, the work advances real-time, image-based ECG analysis and lays groundwork for edge-enabled, explainable arrhythmia detection in home and clinical settings.

Abstract

In a landscape characterized by heightened connectivity and mobility, coupled with a surge in cardiovascular ailments, the imperative to curtail healthcare expenses through remote monitoring of cardiovascular health has become more pronounced. The accurate detection and classification of cardiac arrhythmias are pivotal for diagnosing individuals with heart irregularities. This study underscores the feasibility of employing electrocardiograms (ECG) measurements in the home environment for real-time arrhythmia detection. Presenting a fresh application for arrhythmia detection, this paper leverages the cutting-edge You-Only-Look-Once (YOLO)v8 algorithm to categorize single-lead ECG signals. We introduce a novel loss-modified YOLOv8 model, fine-tuned on the MIT-BIH arrhythmia dataset, enabling real-time continuous monitoring. The obtained results substantiate the efficacy of our approach, with the model attaining an average accuracy of 99.5% and 0.992 mAP@50, and a rapid detection time of 0.002 seconds on an NVIDIA Tesla V100. Our investigation exemplifies the potential of real-time arrhythmia detection, enabling users to visually interpret the model output within the comfort of their homes. Furthermore, this study lays the groundwork for an extension into a real-time explainable AI (XAI) model capable of deployment in the healthcare sector, thereby significantly advancing the realm of healthcare solutions.
Paper Structure (24 sections, 17 equations, 9 figures, 7 tables)

This paper contains 24 sections, 17 equations, 9 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Preprocessing
  4. Generating datasets
  5. You-Only-Look-Once (YOLO)
  6. Backbone
  7. Neck
  8. Head
  9. Loss
  10. Loss function modifications
  11. Classification
  12. Bounding box regression
  13. Training
  14. Evaluation methods
  15. YOLO evaluation metrics
  16. ...and 9 more sections

Figures (9)

  • Figure 1: Flow diagram of the proposed methodology
  • Figure 2: YOLOv8 nano (YOLOv8n) architecture, diagram based on terven2023comprehensiveJocher_YOLO_by_Ultralytics_2023ju2023fracture.
  • Figure 3: Schematic diagram of bounding box parameters where subscripts $gt$, $p$, $i$, and $g$ denote ground truth, predicted, intersected, and smallest enclosed, respectively. $x$ and $y$ are the coordinates of the bounding boxes.
  • Figure 4: Mosaic augmentation for epochs 1-50 (left) and without mosaic augmentation for epochs 51-200 (right).
  • Figure 5: Training and validation results (YOLOv8n + DICF + WIoU).
  • ...and 4 more figures