dwi_skript.Rmd

---
title: "dwi"
author: "Niklas Wulms"
date: "7/15/2020"
output: html_document
editor_options: 
  markdown: 
    wrap: 72
---

Install fsl - take care - you need to set the .bash_profile or .bashrc path variables. Otherwise the script won't work.
Always start R/Rstudio from the Terminal - so that is has access to fsl. Otherwise no processing happens.

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(tidyverse)
library(doParallel)
library(foreach)
library(oro.nifti)
library(neurobase)
library(xml2)
library("XML")


source("dwi_processing/dwi_functions.R")

# parallelization option
cores_not_to_use = 1
topup = "off" # (on/off) sets topup "on" or "off"
end_on_fdt = "yes" # (yes/no) "yes": if you have an error, that files with fdt are not found, turn this on. "no": "_FA" will be applied to the end of the filename.
```





```{r}
# this folder needs a subfolder "sourcedata" containing data in BIDS specification 
# and a folder "params" with "acqparams.txt and index.txt

## tric data
# BIDS_folder <- "data/data_tric/bids" # take care - should end without the last "/"
# subject_id <- "sub-(AD|ADHC|BRP|BRPHC)[:digit:]{5}/"

## bidirect data
# BIDS_folder <- "data/data_bidirect_0_0059375/bids/" # error - unrealistic readout time?
BIDS_folder <- "data/data_bidirect_5_9375/bids/" # error take care - should end without the last "/"

# 0.59 errpr, 0.059 geht?

subject_id <- "sub-[:digit:]{5}/"


BIDS_folder <- normalizePath(BIDS_folder)


# this file is in FSL folder!
b02b0_cnf <- "/usr/local/fsl/src/topup/flirtsch/b02b0.cnf"

source("dwi_processing/filenames/filenames.R", local = TRUE)


```


# Preparing

Calculate index.txt, if it does not exist

```{r}
path_to_folder(index)

# Create Index file if it does not exists
if(!file.exists(index)){
  dwi_length_bvals <- dwi_bval[1] %>% 
    read.table(header = FALSE, sep = "", dec = ".") %>% 
    length()
  
  index_txt <- rep(1, dwi_length_bvals)
  write_lines(x = index_txt, 
              file = index, 
              sep = " ")
  
  index %>% read.table(header = FALSE, sep = "", dec = ".")
}
```

acqparams.txt

this file is sequence and user specific and needs to be added manually!

needs to be in the bids/derived/TRIC_dwi_pipeline/params folder

```{r}
acqparams %>% read.table(header = FALSE, sep = "", dec = ".")

```

## DWI preprocessing

```{r}
source("dwi_processing/methods/0_dwi_preprocessing.R")
```

```{r}
source("dwi_processing/methods/1_dti_processing.R")
```


# TBSS like pipeline

## tbss FA

calculate TBSS for FA in the "3_tbss" folder

```{r}
source("dwi_processing/filenames/filenames_tbss.R")
source("dwi_processing/methods/2_TBSS_FA_processing.R")
```


## tbss non-FA

first copies the folder and content "tbss" from "3_tbss" to "4_nonFA".

Calculate the AD, RD files.  

- AD is L1.
- RD is calculated from L2 and L3

Put AD, RD and MD into folders "AD" or "RD" or "MD".  

And rename them to FA.  

Run TBSS on the whole folder.

Now the output should be in the "stats" folder.

    
```{r}
source("dwi_processing/methods/3_TBSS_nonFA_processing.R")
```


    





# ROI extraction





```{r}
source("dwi_processing/filenames/filenames_extraction.R")
```

```{r}
# inputs
atlas <- "/usr/local/fsl/data/atlases/JHU/JHU-ICBM-labels-1mm.nii.gz" # exact path to atlas
label_file <- "/usr/local/fsl/data/atlases/JHU-labels.xml" # exact path to labels to atlas

atlas_prefix <- "JHU" # custom label of the atlas.
```

```{r}
# whole-brain - unmasked

# frontal, temporal, parietal, occipital


# roi-based
source("dwi_processing/methods/4_md_fa_extraction.R")
```


```{r}
source("dwi_processing/methods/5_readout_FA_MD_RD_AD.R")
```







# GLM Analysis

glmgui: "Glm"

## GLM Design definieren.

```{r}
stats_nonfa_directory <- paste0(output_tbss_nonfa, "/stats")

tbss_all(input_directory = stats_nonfa_directory,
         input_command = "randomise -i all_FA_skeletonised -o tbss -m mean_FA_skeleton_mask -d design.mat -t design.con -n 500 --T2 -V")


randomise_fa_in <- list.files(stats_nonfa_directory, "^tbss", full.names = TRUE)
randomise_fa_out <- str_replace(randomise_fa_in, "nonFA/tbss/stats/tbss", "randomise/FA/tbss") 
path_to_folder(randomise_fa_out)
file.rename(randomise_fa_in, randomise_fa_out)
```

```{r}
tbss_all(input_directory = stats_nonfa_directory,
         input_command = "randomise -i all_MD_skeletonised -o tbss -m mean_FA_skeleton_mask -d design.mat -t design.con -n 500 --T2 -V")

randomise_fa_in <- list.files(stats_nonfa_directory, "^tbss", full.names = TRUE)
randomise_fa_out <- str_replace(randomise_fa_in, "nonFA/tbss/stats/tbss", "randomise/MD/tbss") 
path_to_folder(randomise_fa_out)
file.rename(randomise_fa_in, randomise_fa_out)
```


```{r}
tbss_all(input_directory = stats_nonfa_directory,
         input_command = "randomise -i all_RD_skeletonised -o tbss -m mean_FA_skeleton_mask -d design.mat -t design.con -n 500 --T2 -V")

randomise_fa_in <- list.files(stats_nonfa_directory, "^tbss", full.names = TRUE)
randomise_fa_out <- str_replace(randomise_fa_in, "nonFA/tbss/stats/tbss", "randomise/RD/tbss") 
path_to_folder(randomise_fa_out)
file.rename(randomise_fa_in, randomise_fa_out)
```

```{r}
tbss_all(input_directory = stats_nonfa_directory,
         input_command = "randomise -i all_AD_skeletonised -o tbss -m mean_FA_skeleton_mask -d design.mat -t design.con -n 500 --T2 -V")

randomise_fa_in <- list.files(stats_nonfa_directory, "^tbss", full.names = TRUE)
randomise_fa_out <- str_replace(randomise_fa_in, "nonFA/tbss/stats/tbss", "randomise/AD/tbss") 
path_to_folder(randomise_fa_out)
file.rename(randomise_fa_in, randomise_fa_out)
```




## randomise -i \<4D_input_data\> -o <output_rootname> -d design.mat -t
design.con -m <mask_image> -n 500 -D -T

```{r}
command14 <- paste0("tbss_fill", tbss_clustere_corrp_tstat1,
                    " 0.949",  mean_FA, tbss_clustere_corrp_tstat1_filled)
print(command14)
lapply(command14, system)
```

## tbss_fill tbss_clustere_corrp_tstat1 0.949 mean_FA
tbss_clustere_corrp_tstat1_filled