create_scansheet

#!/usr/bin/env Rscript
library(jsonlite)
library(data.table)

#read in command line arguments.
args <- commandArgs(trailingOnly = FALSE)

scriptpath <- dirname(sub("--file=", "", grep("--file=", args, fixed=TRUE, value=TRUE), fixed=TRUE))
argpos <- grep("--args", args, fixed=TRUE)
if (length(argpos) > 0L) {
   args <- args[(argpos+1):length(args)]
} else {
  args <- c()
}

if (length(args) == 0L) {
  message("create_scansheet expects the root of the MR_Raw directory as the first parameter.\n")
  #printHelp()
  quit(save="no", 1, FALSE)
}

rawdir <- args[1]
#rawdir <- "/gpfs/group/mnh5174/default/PANDAA/MR_Raw"

setwd(rawdir)
alldf <- list()
#make sure we have directories that exist, not symbolic links or other weird things
subjdirs <- unlist(sapply(list.dirs(rawdir, full.names=TRUE, recursive=FALSE), function(d) {
  i <- file.info(d)
  if (!is.na(i$isdir) && i$isdir) { return(d)
  } else { return(NULL)
  }
}))

for (s in subjdirs) {
  setwd(s)
  rundirs <- sort(grep("\\d+_", list.dirs(s, full.names=TRUE, recursive=FALSE), perl=TRUE, value=TRUE))
  for (r in rundirs) {
    setwd(r)
    subjid <- basename(dirname(r))
    run_name <- basename(r)
    jsonfile <- file.path(r, paste0(run_name, ".json"))
    if (!file.exists(jsonfile)) {
      system(paste("dcm2niix -ba n -b o -f", run_name, ".")) #create the json file if missing
    }

    j <- read_json(jsonfile)
    alldf[[paste0(subjid, "_", run_name)]] <- list(ID=subjid, run_name=run_name, location=j$InstitutionName,
      protocol_name=j$ProtocolName, series_number=j$SeriesNumber,
      acq_date_time=j$AcquisitionDateTime, dcm_PatientName=j$PatientName, dcm_PatientID=j$PatientID,
      tr=j$RepetitionTime, echo_time=j$EchoTime, flip_angle=j$FlipAngle,
      slice_thickness=j$SliceThickness, in_plane_voxels=j$BaseResolution,
      phase_encoding_direction=j$PhaseEncodingDirection,
      effective_echo_spacing=j$EffectiveEchoSpacing,
      bandwidth=j$PixelBandwidth)
  }
  
}


#if fields are missing in the json, we will get NULLs. Replace these with NAs to make it friendly to a data.frame structure.
alldf <- lapply(alldf, function(run) {
  lapply(run, function(x) ifelse(is.null(x), NA, x))
})
alldf_bound <- rbindlist(alldf)

write.csv(alldf_bound, file=file.path(rawdir, "scansheet.csv"), col.names=TRUE, row.names=FALSE)