Signal Quality Analysis for Long-Term ECG Monitoring Using a Health Patch in Cardiac Patients

Figures

• 

  Figure 1. Single-lead Vital Signs patch exploded view.

• 

  Figure 2. Single-lead Vital Signs patch based on self-adhesive dry electrodes.

• 

  Figure 3. Cardiac patient during the EST wearing standard 12-lead ECG electrodes and the Vital Signs patch in the 5th intercostal space.

• 

  Figure 4. Summary of the processing steps.

• 

  Figure 5. Example results of preprocessing method to reject irregular QRS complexes. (a) Segmented QRS complexes. (b) Segmented QRS complexes variance distribution and confidence intervals. (c) Accepted segmented QRS complexes. (d) Rejected segmented QRS complexes. Signal colours are arbitrary to keep the segmented QRS complexes identifiable.

• 

  Figure 6. Example protocol timeline and separations used in the analysis. The ∼ symbol indicates the partial full days and daytime.

• 

  Figure 7. Confusion matrices for the SQIs compared using the QDB dataset. (a) Confusion matrix for SQIavg. (b) Confusion matrix for SQIfuzzy. (c) Confusion matrix for SQIhrv. (d) Confusion matrix for SQIQRS.

• 

  Figure 8. Distribution and statistical analysis of ECG quality during daytime and night for all patients.. The explanation of *, **, ***, **** and ns correspond to the definitions of Table 5.

• 

  Figure 9. ECG segments of recording segments by Vital Signs patch in different settings (during the day, night, and exercise) with quality higher and lower than 0.8. (a) ECG segment during the stress test, quality signal >0.8. (b) ECG segment during the day, quality signal >0.8. (c) ECG segment during the night, quality signal >0.8. (d) ECG segment during the stress test, quality signal <0.8. (e) ECG segment during the day, quality signal <0.8. (f) ECG segment during the night, quality signal <0.8.

• 

  Figure 10. Data quality over time for all patients. Some patients have fewer days due to data loss.