Load Data by Signia Hearing Aids

This example shows you how to load data recorded with Signia hearing aids. Please note, that the privat python package signialib is required for running this example. Furthermore, if you want to use the loading functionalities your data needs to be *.mat files with the same data structure as the *.mat files in the example_data folder.

Getting some example data

For this we take some example data that contains regular walking movements. data_path is the local path of the directory containing the recorded data. Note: The directory should any contain a single recoding (maximum two *.mat files, for each ear one)

# from eargait.event_detection import DiaoAdaptedEventDetection
# from eargait.preprocessing import align_gravity_and_convert_ear_to_ebf, convert_ear_to_ebf, load

from pathlib import Path

from eargait.utils.example_data import get_mat_example_data_path, get_txt_example_data_path

# data directory
data_path = get_mat_example_data_path()
data_path

PosixPath('/home/docs/checkouts/readthedocs.org/user_builds/eargait/checkouts/latest/example_data/mat_files/fast')

Loading an individual Dataset

Note: This method loads a single *.mat file. If the recording constitutes a *.mat file for both ears, it’s not recommended to use this function but to load a Session as shown in the next section.

from signialib import Dataset

dataset = Dataset.from_mat_file(data_path.joinpath("data_left.mat"))
# not recommended, recommended to use Session.from_folder_path(data_path) instead.

Loading the data from mat files

from signialib import Session

session = Session.from_folder_path(data_path)

Loading the data from txt files

For txt files a different loading function is required. The function from_file_path expects a specific file path of either a txt file or a single mat file, e.g. user/example.txt or user/example.mat. All other functionalities (alignment, resampling etc., see below) for sessions are the same for all file types.

from signialib import Session

data_path_txt = get_txt_example_data_path()
session_txt = Session.from_file_path(data_path_txt)

Get information about session

session.info

# should display something like this:
from eargait.utils.example_data import plot_image

IMAGE = plot_image(data_path.parent.parent.parent.joinpath("docs/_static/example_screenshots/session_info.jpeg"))

Get information about sensors

Two examples for how to access further information on e.g. sensor_ids or sampling rate

session.info.sensor_id

('001', '002')

session.info.sampling_rate_hz

(200.0, 200.0)

Align and calibrate data

The two recordings of the left and right ear might slightly defer in the number of samples due to unavoidable inaccuracies in the sample rate. The alignment corrects these small differences. For the beginning set skip_calibration=True.

aligned_session = session.align_calib_resample(skip_calibration=True)

Note: The calibration_files can be skipped using the skip_calibration parameter. However, a calibration_files for IMU sensors to align sensor axes to the individual sensor case is recommended using the calibration_files method proposed by Ferraris et al. [1]. The imucal package [2] provides a code and a detailed description of the calibration_files.

[1] [Ferraris et al., Calibration of three-axial rate gyros without angular velocity standards, Sensors and Actuators A: Physical, 1994]

[2] [(https://imucal.readthedocs.io)]

Resampling

resampled_session = session.align_calib_resample(skip_calibration=True, resample_rate_hz=50)
resampled_session.info.sampling_rate_hz

(50.0, 50.0)

Get single dataset

Either by position (ha_left, ha_right) or by ID

left_dataset = resampled_session.get_dataset_by_position("ha_left")
dataset_1 = resampled_session.get_dataset_by_id("001")

Getting a single datastream

Either acceleration or gyroscope data. Furthermore, the datastream can be directly extracted as pd.DataFrame

acc = left_dataset.acc
acc_df = left_dataset.acc.data_as_df()
gyro = left_dataset.gyro
gyro_df = left_dataset.gyro.data_as_df()

Gravity Alignment and Conversion to medical coordinate system (only for singia hearing aids!!)

This method can be used to align the data to gravity. The method is only available for Signia hearing aids, because orientation of IMU within world frame is known. For further information on the different coordinate systems please refer to the documentation Coordinate Systems

from eargait.preprocessing import align_gravity_and_convert_ear_to_ebf
from eargait.utils import StaticWindowGravityAlignment, TrimMeanGravityAlignment

gravity_method = StaticWindowGravityAlignment(resampled_session.info.sampling_rate_hz[0])
ear_data = align_gravity_and_convert_ear_to_ebf(session, gravity_method)

Gyroscope signal is used for gravity alignment.
Gyroscope signal is used for gravity alignment.

Data visualization

Small example.

import matplotlib.pyplot as plt

fig, axes = plt.subplots(2, 1, sharex=True)
acc = aligned_session.get_dataset_by_position("ha_left").acc.data_as_df()
axes[0].plot(acc, label=acc.columns)
axes[0].set_ylabel("Acceleration [m/s$^2$]")
gyro = aligned_session.get_dataset_by_position("ha_left").gyro.data_as_df()
axes[1].plot(gyro, label=gyro.columns)
axes[1].set_ylabel("Gyroscope [°/s]")
axes[1].set_xlabel("Time [s]")
plt.suptitle("Left Hearing Aid Data")
plt.legend()
plt.tight_layout()
plt.show()
k = 1

Estimated memory usage: 46 MB