⚡ electrical engineering

Textile Electrode Size and Placement Influence on Electrocardiography Monitoring

This study demonstrates that while textile electrode size does not significantly impact ECG signal quality, placement location is critical, with positions under the left shoulder blade on the back and under the left nipple on the chest yielding optimal results that often surpass conventional Ag/AgCl electrodes during physical activity.

Original authors: Martin Lajdolf, Radek Martinek, Rene Jaros

Published 2026-06-26

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Martin Lajdolf, Radek Martinek, Rene Jaros

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine your heart is a tiny, rhythmic drummer inside your chest, sending out electrical signals that tell doctors how you're doing. Usually, to hear this drummer clearly, doctors stick sticky, gel-covered pads (like temporary tattoos) to your skin. But these can be itchy, uncomfortable, and fall off if you move too much.

This research paper asks a simple question: Can we make better "listening pads" out of fabric (textiles) that feel like normal clothes, and does it matter how big those pads are or exactly where we put them?

Here is the breakdown of their experiment and findings, explained simply:

The Experiment: The "Fabric Patch" Test

The researchers built special fabric electrodes in three different sizes:

Small: Like a postage stamp (2x2 cm).
Medium: Like a large coaster (4x4 cm).
Large: Like a small postcard (6x6 cm).

They tested these patches on 15 volunteers in two main ways:

Resting: Just sitting or standing still.
Moving: Doing physical activities like walking, lifting weights, and squatting.

They placed these fabric patches in six different spots: three on the front of the chest and three on the back. For every spot, they also stuck a standard "gold standard" medical patch right next to it to compare the signals. Think of it as having a fabric microphone and a professional studio microphone recording the same singer at the same time to see which one sounds better.

The Big Discovery 1: Size Doesn't Matter Much

The Analogy: Imagine trying to hear a conversation in a noisy room. Does it matter if you use a small ear or a giant ear? In this study, the answer was no.

The researchers found that whether the fabric patch was small, medium, or large, the quality of the heart signal was almost exactly the same.

The Result: You don't need a giant patch to get a good reading. A smaller patch works just as well.
Why this is cool: Smaller patches mean you could potentially put more of them on a shirt to monitor different things at once, or make the clothing less bulky.

The Big Discovery 2: Location is Everything

The Analogy: Think of the heart's signal like a radio station. If you are standing right next to the radio tower (the heart), the signal is crystal clear. If you stand 10 miles away, it's full of static.

The study found that where you put the patch mattered way more than how big it was.

Front vs. Back: Surprisingly, the patches on the back of the body gave much clearer signals than the ones on the front. The researchers suggest this might be because there is less fat and muscle between the heart and the back skin, making the "radio signal" stronger.
Left vs. Right: On both the front and the back, the spots closest to the heart (under the left side) worked the best.
- Best Spot on Chest: Under the left nipple.
- Best Spot on Back: Under the left shoulder blade.

The Big Discovery 3: Moving Around (The "Jitter" Test)

When the volunteers started moving (walking, lifting weights), the signals usually get messy (like static on a radio when you drive through a tunnel).

The Result: The fabric patches were actually better than the sticky medical patches at catching the heartbeats while walking or doing light exercises.
The Exception: When people were doing heavy weightlifting (dumbbells) or deep squats, the standard sticky patches held up slightly better, but the fabric patches were still very close in performance.

The "Secret Weapon": The AI Brain

To make sure they were counting the heartbeats correctly, the researchers didn't just look at the squiggly lines; they used a Deep Neural Network (a type of artificial intelligence).

The Analogy: Imagine a human trying to count heartbeats in a noisy crowd. It's hard. But imagine a super-smart robot that has listened to thousands of heartbeats. It can instantly spot the "beat" even if there is noise or weird shapes in the signal.
The AI helped confirm that the fabric patches were accurate enough to be trusted, even when the person was moving.

The Bottom Line

If you want to build a smart shirt to monitor your heart:

Don't worry about making the patches huge. Small ones work just as well.
Put the patches on the back. Specifically, under the left shoulder blade, or on the front under the left nipple.
It works while you move. These fabric patches are great for active people, often beating the old sticky pads during walking.

In short, the paper proves that we can replace uncomfortable, sticky medical pads with comfortable fabric ones, as long as we put them in the right spot. The size of the patch is a minor detail; the location is the real key to a clear signal.

Technical Summary: Textile Electrode Size and Placement Influence on Electrocardiography Monitoring

Problem Statement
Long-term monitoring of cardiac electrical activity (ECG) via wearable devices is increasingly prevalent. However, conventional silver/silver chloride (Ag/AgCl) electrodes present significant limitations, including user discomfort, potential allergic reactions, skin irritation, and susceptibility to motion artifacts. While textile electrodes offer a promising alternative due to their reusability, skin friendliness, and ease of integration into clothing, their signal quality is heavily dependent on specific design parameters. The authors identify a gap in the literature regarding the systematic evaluation of how electrode contact area size and specific body placement influence signal quality and R-peak detection accuracy, particularly under varying conditions of rest and physical activity.

Methodology
The study employed a robust experimental design involving 30 volunteers of varying ages, genders, and somatotypes. The methodology was divided into two primary phases:

Signal Quality Assessment (Rest and Activity):
- Electrode Design: Three sizes of textile electrodes were manufactured using Silitex conductive fabric (Shieldex) with memory foam padding to ensure skin contact. The sizes were 2×2 cm, 4×4 cm, and 6×6 cm.
- Placement: Six distinct locations were tested: three on the chest (below the right nipple, right of the sternum, middle of the sternum) and three on the back (below the right shoulder blade, spinal groove, below the left shoulder blade). Each textile electrode was paired with a reference Ag/AgCl electrode placed immediately below it.
- Data Acquisition: Signals were recorded using a commercial BTL Flexi 12 ECG device at 500 Hz. Raw data was processed using 2nd-order IIR filters (high-pass at 0.5 Hz, low-pass at 100 Hz, and band-stop at 49–51 Hz).
- Evaluation Metrics: Signal quality was assessed by comparing representative cardiac periods (RCP) derived from textile signals against reference signals using Pearson's correlation coefficient ( $R$ ) and Mean Squared Error (MSE).
R-Peak Detection Accuracy:
- Model: A Deep Neural Network (DNN) was designed for R-peak detection. The architecture features three parallel 1D convolutional branches with varying kernel sizes to capture different QRS scales, a bidirectional LSTM layer for sequential rhythm information, and dual output heads (a detection head with a Gaussian probability map and a regression head for precise localization).
- Training: The model was trained on approximately 1,200 annotated 6-minute ECG recordings from both textile and reference electrodes under laboratory conditions.
- Validation: The F1 score (harmonic mean of precision and recall) was used to evaluate detection accuracy. A supplementary dataset involving 15 additional volunteers performing four physical activities (walking at 4 km/h and 5 km/h, dumbbell exercises, and squats) was used to test performance under motion artifacts.

Key Contributions

Systematic Size Evaluation: The study tested three distinct textile electrode sizes (2×2, 4×4, and 6×6 cm) to determine the impact of contact area on signal fidelity.
Comprehensive Placement Analysis: The research evaluated six specific body locations (three anterior, three posterior) to map the influence of anatomical position on ECG quality.
Advanced Detection Model: The development and application of a custom DNN model for high-precision R-peak detection, capable of handling diverse signal morphologies.
Robust Dataset: The generation of a dataset comprising measurements from 30 volunteers, including specific subsets for physical activity analysis, providing a broader sample size than many comparable studies.
Multi-Metric Evaluation: A holistic assessment using correlation ( $R$ ), MSE, and F1 scores to evaluate both signal morphology and detection algorithm performance.

Results

Electrode Size: Statistical analysis (Kruskal-Wallis test) revealed no statistically significant difference in signal quality based on electrode size. While the smallest electrodes (2×2 cm) occasionally showed slightly lower variability and higher median correlation values, the differences between sizes were minimal.
Electrode Placement: Placement had a statistically significant impact on signal quality.
- Chest vs. Back: Signals recorded from the back yielded significantly better results (higher $R$ , lower MSE) than those from the chest.
- Specific Locations: On the chest, the position below the left nipple yielded the best results. On the back, the position below the left shoulder blade achieved the highest quality, with data clustering near perfect correlation ( $R \approx 1$ ) and zero error.
- Distance Correlation: Signal quality improved as the electrode position moved closer to the heart (left side), with the right side showing the lowest quality.
R-Peak Detection (F1 Score):
- Under resting conditions, F1 scores were generally high (>98%) for both textile and reference electrodes across most positions.
- During physical activity, textile electrodes at the V4 position demonstrated better R-peak detection stability than reference electrodes for walking (4 km/h and 5 km/h) and dumbbell exercises.
- The only exception was weight training (dumbbells/squats), where reference electrodes generally performed slightly better or comparably, particularly at the V6 position.
- At V5 and V6 positions, differences between textile and reference electrodes were minimal across all activities.

Significance and Claims
The paper concludes that while electrode size is not a critical factor for signal quality, placement is the dominant variable. The study claims that textile electrodes can serve as a functional replacement for conventional Ag/AgCl systems, provided they are placed optimally (specifically on the back, near the left shoulder blade, or on the chest near the left nipple).

The authors emphasize that the superior performance of smaller electrodes (2×2 cm) in terms of consistency suggests potential for minimizing contact areas in wearable designs. This could allow for higher sensor density in wearable garments, enabling the simultaneous monitoring of multiple biological signals (e.g., ECG and EMG) without compromising comfort. The study asserts that the proposed DNN model effectively handles the morphological diversity of textile ECG signals, validating the feasibility of using deep learning for artifact-prone wearable data.

The work positions itself as a necessary step in the state of the art, addressing the lack of comprehensive studies on the combined effects of size, placement, and motion on textile ECG, and provides a validated dataset and methodology for future research in wearable cardiac monitoring.