PPG Pre-processing and SQI Calculations

Trimming, Filtering, Segmentation and SQI Calculations

 # Code Adapted from:
 # https://github.com/meta00/vital_sqi/blob/main/examples/SQI_pipeline_PPG.ipynb
 # AND
 # https://meta00.github.io/vital_sqi/_examples/others/plot_read_signal.html#sphx-glr-examples-others-plot-read-signal-py

Libraries

 # Generic
 import os
 import numpy as np
 import pandas as pd
 import matplotlib.pyplot as plt

 # Scipy
 from scipy.stats import skew
 from scipy.stats import kurtosis

  # vitalSQI
 from vital_sqi.data.signal_io import PPG_reader
 import vital_sqi.highlevel_functions.highlevel as sqi_hl
 import vital_sqi.data.segment_split as sqi_sg
 from vital_sqi.common.rpeak_detection import PeakDetector

Out:

Importing the dtw module. When using in academic works please cite:
  T. Giorgino. Computing and Visualizing Dynamic Time Warping Alignments in R: The dtw Package.
  J. Stat. Soft., doi:10.18637/jss.v031.i07.

Load data

 # Filepath

 filepath = r'..\..\..\..\OUCRU\01NVa_Dengue\Adults\01NVa-003-2001\PPG'
 filename = r'01NVa-003-2001 Smartcare.csv'


 #defining constants
 trim_amount = 120
 hp_filt_params = (1, 1) #(Hz, order)
 lp_filt_params = (20, 4) #(Hz, order)
 filter_type =  'butter'
 segment_length = 30
 sampling_rate = 100

 #readind PPG data
 data = PPG_reader(os.path.join(filepath, filename),
     signal_idx=["PLETH", "IR_ADC"],
     timestamp_idx=["TIMESTAMP_MS"],
     info_idx=["SPO2_PCT","PULSE_BPM","PERFUSION_INDEX"], sampling_rate=sampling_rate)


 print(data)
 print(data.signals)
 print(data.sampling_rate)
 print(data.start_datetime)
 print(data.wave_type)
 print(data.signals.shape)


 plot_range = np.arange(0,1000,1)
 fig, ax = plt.subplots()
 ax.plot(plot_range, data.signals[0][0:1000])

Out:

<vital_sqi.data.signal_sqi_class.SignalSQI object at 0x000001AC953D1670>
[[ 36033  40919  45715 ...  31117  30761  30261]
 [216524 216470 216415 ... 184296 184554 184828]]
100
None
ppg
(2, 5626377)

[<matplotlib.lines.Line2D object at 0x000001AC96431BB0>]

Filtering, Trimming and Segmentation

 #Saving a copy of the original Sigal

 original_signal = np.copy(data.signals.T[trim_amount*sampling_rate:-trim_amount*sampling_rate].T)
 original = pd.DataFrame(original_signal)
 print("Raw Signals Shape: ", original.shape)


 #Signal  Filtering

 data.update_signal(sqi_hl.signal_preprocess(signal_channel=data.signals, hp_cutoff_order=hp_filt_params, lp_cutoff_order=lp_filt_params, trim_amount=trim_amount, filter_type=filter_type, sampling_rate=sampling_rate))
 signals = pd.DataFrame(data.signals)
 print("Filtered Signals Shape: ", signals.shape)

 fig, ax = plt.subplots()
 ax.plot(plot_range, data.signals[0][0:1000])


 # Signal Segmenation

 data.update_segment_indices(sqi_sg.generate_segment_idx(segment_length=segment_length, sampling_rate=sampling_rate, signal_array=data.signals))
 print("Data Segments Shape: ", data.segments.shape)
 segments= pd.DataFrame(data.segments)

Out:

Raw Signals Shape:  (2, 5602377)
Filtered Signals Shape:  (2, 5602377)
Data Segments Shape:  (2, 1868)

Peak Detection

 #Peak Detector
 peak_list, trough_list = PeakDetector().ppg_detector(signals.iloc[0,segments.iloc[0,0]:segments.iloc[0,1]], 7)

 plot_range = np.arange(0,3000,1)
 fig, ax = plt.subplots()
 ax.plot(plot_range, signals.iloc[0, segments.iloc[0,0]:segments.iloc[0,1]])
 if len(peak_list)!=0:
     ax.scatter(peak_list,signals.iloc[0,peak_list],color="r",marker="v")
 if len(trough_list)!=0:
     ax.scatter(trough_list,signals.iloc[0,trough_list],color="b",marker="v")


 #Single period plot
 fig, ax = plt.subplots()
 ax.plot(signals.iloc[0,trough_list[0]:trough_list[1]])


 print(data.signals[0])
 print(data.signals.shape)
 print(segments.shape)
 print(data.segments.shape)
 print(original_signal.shape)
 print(peak_list.shape)
 print(trough_list.shape)

• 
• 

Out:

[ 142  821 2090 ... 1555 1462 1302]
(2, 5602377)
(2, 1868)
(2, 1868)
(2, 5602377)
(71,)
(71,)

SQI Calculation

 #THE FUNCTION BELOW OUTPUTS AN ERROR, NEEDS TO BE EDITED SO THAT if signals are none, they are replaced with NaN

 #computed_sqi = sqi_hl.compute_all_SQI(signal=data.signals[0], segments=data.segments[0], raw_signal=original_signal[0], primary_peakdet=7, secondary_peakdet=6, template_type=0)

 '''
 ERROR THAT COMES UP:

 raise Exception("Not enough peaks in the signal to generate per beat SQI")
 Exception: Not enough peaks in the signal to generate per beat

 THIS COMES UP FOR PLETH (index 0) BUT NOT FOR IR_ADC (index 1)

 '''


 '''
 computed_sqi = pd.DataFrame(computed_sqi)


 print(computed_sqi[2].keys())
 print(data.segments.shape)
 print(peak_list.shape)
 print(trough_list.shape)
 print(computed_sqi[1873]['snr'])
 '''

Out:

"\ncomputed_sqi = pd.DataFrame(computed_sqi)\n\n\nprint(computed_sqi[2].keys())\nprint(data.segments.shape)\nprint(peak_list.shape)\nprint(trough_list.shape)\nprint(computed_sqi[1873]['snr'])\n"

Segment Rejection

 #AFTER COMPUTING SQIs REJECT USING CRITERIA IS THE NEXT STEP.












 plt.show()