Molecool:Method

From OpenWetWare
Jump to navigationJump to search

Molecool HRV Logo.png

Home        Background Research        Clinical Application        Method        Social Practice        Result        Journal        Lab Members       


General Plan in Data Extraction and Processing

To achieve clinical purposes using HRV, it is important to convert the ECG diagrams into quantifiable measures. Therefore, data extraction from ECG and data processing to obtain desired quantities are the essential parts of this project. The complete plan is shown in the block diagram below.


Block diagram data.png


Step 1: ECG Data Acquisition from Database
Step 2: Preprocessing by applying filters to correct baseline and clear noise
Step 3: R peak detection through Pan Tompkins algorithm
Step 4.1: Perform time-domain analysis by calculating the parameters
Step 4.2: Perform frequency-domain analysis by Fast Fourier Transform in high-frequency and low-frequency ranges
Step 5: Achieve diabetic neuropathy monitoring through data classification

Common Methods in HRV Data Processing

Pre-processing

R-peak Detection

Database

Database Introduction

The dataset provides data from type 2 diabetic and control patients. Two groups of 70 patients aged 50-85 years old each, one control group and an experimental group with the disease, form the dataset. The methods used for data acquisition were significant when selecting it as a potential database. The diabetes patients had had type 2 DM for more than a year. The patients from the control group were carefully chosen, following both inclusive and exclusive criteria. Both are summarised in a protocol attached alongside the database, which can be found here: https://physionet.org/content/cerebral-perfusion-diabetes/1.0.0/GE-75_protocol.pdf.
The ECG recordings contain values for two full days, giving us data from a 24h (DAY 1) monitoring. Hence, it enables us to double-check if external parameters can have an impact on the ECG recordings. e.g. day time compared to night time, signal changes after ingesting food, mood fluctuations, etc. Finally, as HRV extraction can only consist of ten-minute intervals, we have different data sets to prove a potential conclusion in future scheme stages.

Database Description

  • 6 Controls, 45 Diabetic – 51 in total
  • 2 readings per patient (24 hr/walking exercise)
    • Ecg_1: CH1 V5/V6-L clavicle
    • Ecg_2: CH1 V1/V2-L clavicle
    • Also have ECG for sit to stand + head-up-tilt tests
  • Sampled at 1000Hz
  • Separated into:
    • Small (>10Mb, <50Mb. 39 in total)
    • Large (>800Mb, largest is 911Mb. 11 in total)
    • one of the samples is 140Mb

Our Methodology of HRV Data Processing

Pre-processing

R-peak Detection