MMGMarkdown.Rmd

---
title: "MMGExample"
author: "Alice Rueda"
date: "February 19, 2019"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

## MMG Signal Processing
This file reads the MMG signal from *.txt files and display the raw signal before processing. It provides the following functions:
- Read the raw data file
- Plot the raw and processed signal
- Pre-processing the necessary signal including detrendng, filtering, and segment the signals into activity windows with use of PULSE signals
- Extract features from the clean segmented signals
```{r packages}
source("MMGLib.R")
source("PlotSignals.R")
source("TREMSENToolbox.R")

#subjectCode: get it from the directory
#pathData ="D:/Adriano/OneDrive/Projetos/2017-Brazil-Canada/DATA/EJRR/09Nov2018/EMG/Excel"
pathData = "/media/alice/DATA/2019 Adriano/2019 Adriano MMG/Data/Coletas 14_11/MMG_Tremsen/COLETAS 14_11"
# defining subject and filetype
#subjectCode = "RCS"
signalType = "MMG"
fileType = "txt"
```

## Reading MMG file

The data files are stored in txt files with tab as separators. This part lists all files with *.txt in the data file directory.

```{r Read File}
# list all relevant files
#pathFile <- paste(pathData,subjectCode,'/', sep='')
filename <- sort(list.files(path=pathData, full.names=TRUE, pattern="*.txt"))
#filename <- basename(filename)

i <- 1

df <- LoadTREMSENFile(filename[i])
```
## Visualize Raw Signal 

We are only interested in the first 2 sets of accelerometers. Gyroscope signal and magnetometers are not of our interest. This part plots the triaxial signals from the 2 accelerometers.


```{r Visualize Raw Signal}

# Calling utility function from PlotSignal.R to plot the raw signal

ggplotAccelerometer(df)


```



```{r Signal Preprocessing  1}

# Detrend and remove DC offset --------------------------------------------

df.nonlineardetrended <- nonLineardetrendTremsenData(df) #Remove both linear and nonlinear trends

ggplotAccelerometer(df.nonlineardetrended)

```

```{r Signal Processing 2}

# Event detection  ---------------------------------------------------

# Using detrend dataframe for event detection
dfdetrended<-df.nonlineardetrended

eventDetect = detectMMGEvent(dfdetrended, percentage = 0.05, PLOT=TRUE)

scaling = 0.5
channel = dfdetrended$X.A1.X.
dfplot1<-data.frame(time = dfdetrended$X.Time., channel, dfdetrended$X.PULSE*max(channel)*scaling, eventDetect*max(channel)*scaling)
dygraph(dfplot1, main = "Accelerometer x-axis") %>%
  dyAxis("x", label="Time (s)") %>% 
  dyAxis("y", label="Angular velocity") %>%
  dyOptions(colors = c('black', 'blue', 'red')) %>% 
  dyLegend(width = 130) %>%
  dyRangeSelector() 
  

```
```{r Lowpass Filter}

filteredLP <- butterLowPassFilter(channel, cutoff = 2, Fs = 50)   # Motion artifact
filteredHP <- butterHighPassFilter(channel, cutoff = 2, Fs = 50)  # MMG signals
residual <- channel - filteredLP        # Just to show that the MMG signal is the same as subtracting the motion artifacts from the original signal

#dfplot<-data.frame(time = df$X.Time., eventDetect*max(channel), filtered, channel, (channel-filtered))
dfplot<-data.frame(time = dfdetrended$X.Time., channel, filteredLP, eventDetect*max(channel))
dygraph(dfplot, main = 'With Low Pass Filter')%>%
  dyOptions(colors = c('black', 'blue', 'red')) %>%
  dyRangeSelector()


#dfplot<-data.frame(time = df$X.Time., eventDetect*max(channel), filtered, channel, (channel-filtered))
dfplot<-data.frame(time = dfdetrended$X.Time., residual, filteredHP, eventDetect*max(residual))
dygraph(dfplot, main = 'With Low Pass Filter')%>%
  dyOptions(colors = c('black', 'blue', 'red')) %>%
  dyRangeSelector()


differences <- filteredHP - residual

#dfplot<-data.frame(time = df$X.Time., eventDetect*max(channel), filtered, channel, (channel-filtered))
dfplot<-data.frame(time = dfdetrended$X.Time., differences, eventDetect*max(residual))
dygraph(dfplot, main = 'With Low Pass Filter')%>%
  dyOptions(colors = c('black', 'blue', 'red')) %>%
  dyRangeSelector()



```



```{r Segmentation}
# Partition signal -----------------------------------------------
sep <- findSep(eventDetect)


partChannel <- partitionChannel(filteredHP,sep)
partOrignal <- partitionChannel(channel,sep)
numSegment <- size(partChannel,2)  # The activities are stored in even number list of partChannel

pinchMMG <- partChannel$'2'
pinchOrignal <- partOrignal$'2'

handMMG <- partChannel$'4'
handOrignal <- partOrignal$'4'

supMMG <- partChannel$'6'
supOrignal <- partOrignal$'6'

flexMMG <- partChannel$'8'
flexOrignal <- partOrignal$'8'


```