# tOLIet: Single-lead Thigh-based Electrocardiography Using Polimeric Dry Electrodes
This repository contains a dataset of ECG signals acquired using an intelligent toilet system equipped with electrodes of various textures, as well as reference ECG recordings obtained from clinical-grade 12-lead systems. The data is intended for research purposes, enabling the development and validation of algorithms for ECG-based classification, recognition, and biometric identification.
---
## Dataset Structure
```
.
├── ECG_EXP/ # Experimental ECG signals from toilet sensors
│ ├── 001.txt
│ ├── 002.txt
│ └── ...
├── ECG_REF/ # Reference clinical ECG signals (12 leads)
│ ├── 001.XML
│ ├── 002.XML
│ └── ...
├── Script/ # Script
│ ├──read_ecg_data.py # Script to read and preprocess experimental ECG
│ ├──read_ref_data.py # Script to read and extract reference ECG
│ ├──data_analysis.py # Wave detection (P, Q, R, S, T)
├── DataSet.csv # Subject metadata
└── README.md # This file
```
---
## File Descriptions
### `DataSet.csv`
Contains metadata for each subject:
- **ID**: Unique subject identifier
- **Age**: Age (years)
- **Weight**: Weight (kg)
- **Height**: Height (cm)
- **Gender**: M or F
- **Observations_field**: Clinical or diagnostic notes
### `ECG_EXP/`
Contains `.txt` files with experimental ECG data collected using four different electrode textures:
- **A1**: Flat texture
- **A2**: Sinusoidal texture
- **A3**: Pyramidal texture
- **A4**: Trapezoidal texture
Each row corresponds to a time sample. Signals are converted from raw values to millivolts using `read_ecg_data.py`.
### `ECG_REF/`
Contains `.XML` files with 10-second 12-lead ECG reference recordings:
- Acquisition timestamp
- Median heartbeat templates (per lead)
- Raw signals (per lead)
- Unit: microvolts (µV)
---
## Main Scripts
### `read_ecg_data.py`
- Loads experimental `.txt` signals
- Converts raw ADC values to mV
- Returns subject data and signals from sensors A1–A4
### `read_ref_data.py`
- Parses `.XML` files
- Extracts acquisition time, lead names, heartbeat templates, and 10-second raw signals
### `data_analysis.py`
- Detects P, Q, R, S, and T wave positions based on template morphology
- Assumes R peak is always located at index 200 within templates
---
## Usage Instructions
1. **Install required dependencies:**
```bash
pip install biosppy numpy pandas xmltodict scipy
```
2. **Read an experimental ECG signal:**
```python
import read_ecg_data as red
ecg_data = red.read_ecg("ECG_EXP/001.txt", result_from_dataset)
```
3. **Read a reference ECG file:**
```python
import read_ref_data as rrd
ref_data = rrd.read_ref("ECG_REF/001.XML", "001")
```
4. **Extract waveform peak positions:**
```python
import data_analysis as da
peaks = da.getAllPeaks(ecg_proc)
```
---
## Notes
- Templates assume the R peak is always centered at index 200.
- Experimental ECG signals may require filtering due to noise.
- Units:
- Experimental ECG: converted to **mV**
- Clinical reference ECG: recorded in **µV**
---
## Contact
For questions, collaboration opportunities, or additional information, please contact:
- **Name**: Aline Santos Silva
- **Email**: [aline.s.silva@inesctec.pt]
- **Affiliations**:
INESC TEC – Institute for Systems and Computer Engineering, Technology and Science
Faculty of Engineering, University of Porto (FEUP)
IT – Instituto de Telecomunicações
---
## License
This dataset is provided strictly for research purposes and may be subject to ethical approval. Redistribution is not permitted without explicit authorization from the authors.