dcm_anonymize.bash

#!/bin/bash
#
# dcm_anonymize.bash
#
# Copyright 2008 Rudolph Pienaar
# Massachusetts General Hospital
#
# GPL v2
#

# "include" the set of common script functions
source common.bash

declare -i Gi_verbose=0
declare -i Gb_useExpertOptions=0
declare -i Gb_useOverrideOut=0
declare -i Gb_forceStage=1

G_LOGDIR="-x"
G_OUTDIR="/space/kaos/5/users/dicom/postproc"
G_OUTPREFIX="-x"
G_DIRSUFFIX=""
G_OUTPREFIX="anon-"
G_DICOMINPUTDIR="-x"
G_UBUNTUJAVA=""
G_MATLAB="/usr/pubsw/packages/matlab/new/bin/matlab"
G_STAGES="1"

G_SYNOPSIS="

 NAME

	dcm_anonymize.bash

 SYNOPSIS

	dcm_anonymize.bash	-D <dicomInputDir>                      \\
                                [-v <verbosity>]  			\\
				[-O <experimentTopDir>]			\\
				[-R <DIRsuffix>				\\
                                [-o <outputRunDir>]                     \\ 
                                [-p <outputPrefix>]

 DESCRIPTION

	'dcm_anonymize.bash' accepts an input directory containing DICOM
	files and anonymizes the data. Output directory is the standard
	postproc stream.

 ARGUMENTS

	-v <level> (optional)
	Verbosity level.

        -D <dicomInputDir>
        The directory containing DICOM files for a particular study.

        -O <experimentTopDir> (optional) (Default: $G_OUTDIR)
        The root directory node that contains the outputs of a particular
        registration run. Each run is stored in its own directory.
	
	If this is specified on the command line, then output from the 
	anonymization will be written to this directory.

        -o <outputRunDir> 
        A descriptive name prefixed to each volume generated by a registration
        run.

       -R <DIRsuffix> (Optional)
        Appends <DIRsuffix> to the postproc/<MRID> as well as <logDir>. Since
        multiple studies on the same patient can in principle have the same
        MRID, interference can result in some of the log files and source
        data. By adding this <DIRsuffix>, different analyses on the same MRID
        can be cleanly separated.

        -p <outputPrefix> (Optional)
        If specified, prefix all generated output files with <outputPrefix>.

 PRECONDITIONS
	
	o nde
	o MatLAB -- the actual anonymization is performed in MatLAB. This
	  script is merely a thin wrapper that drives the MatLAB process.

 POSTCONDITIONS

	o Output anonymized volumes are stored in:

              <experimentTopDir>/<outputRunDir><DIRsuffix>

          and each file is prefixed by 'anon-'. If an <experimentTopDir>
	  has been explicitly set, then output files are stored directly in
	  <experimentTopDir>.

 HISTORY

	24 March 2009
	o Initial design and coding.

"

###\\\
# Globals are in capital letters. Immutable globals are prefixed by 'G'.
###///
G_SELF=`basename $0`
G_PID=$$

# Actions
A_fileCheck="checking for a required file dependency"
A_noDicomDir="checking on input DICOM directory"
A_noOutRunDir="checking on output run directory"
A_noDicomDirArg="checking on -d <dicomInputDir> argument"
A_noExpDir="checking on the output root directory"
A_metaLog="checking the meta log file"
A_dependencyStage="checking for a required dependency from an earlier stage"
A_stageRun="running a stage in the processing pipeline"

# Error messages
EM_fileCheck="it seems that a dependency is missing."
EM_noDicomDir="I couldn't access the input DICOM dir. Does it exist?"
EM_noOutRunDir="I couldn't access the output run dir. Does it exist?"
EM_noDicomDirArg="it seems as though you didn't specify a -D <dicomInputDir>."
EM_noExpDir="I couldn't find the <expDir>."
EM_metaLog="it seems as though this stage has already run.\n\tYou can force execution with a '-F'"
EM_dependencyStage="it seems that a stage dependency is missing."
EM_stageRun="I encountered an error processing this stage."

# Error codes
EC_fileCheck=1
EC_dependencyStage=2
EC_stageRun=30
EC_noDicomDir=50
EC_noOutRunDir=54
EC_noDicomDirArg=51
EC_noExpDir=23
EC_metaLog=80

# Defaults
D_whatever=

function matlab_scriptCreate
{
    local SCRIPT=$1
    
    cat > $SCRIPT <<-end-of-script
    function [c] 	= dcm_anonymize_drive()
        c	= dcm_anonymize();
        str_in	= '$G_DICOMINPUTDIR';
        str_out	= '$OUTDIR';
        c	= set(c, 'dicomInputDir',	str_in);
        c	= set(c, 'dicomOutputDir',	str_out);
        c	= set(c, 'verbosity',		10);
        c	= set(c, 'b_newSeries',		0);
        c	= set(c, 'anonPrefix',		'$G_OUTPREFIX');
        c	= set(c, 'keep', 		'reset');
        c	= set(c, 'keep', 		'PatientAge');
        c	= set(c, 'keep', 		'PatientSex');       
        c	= set(c, 'keep', 		'StudyDescription');
        c	= set(c, 'keep', 		'SeriesDescription');
        c	= set(c, 'keep', 		'ProtocolName');
	
        c	= run(c);
    end
end-of-script

}

###\\\
# Function definitions
###///


###\\\
# Process command options
###///

while getopts D:Ev:O:o:p:t:R: option ; do 
	case "$option"
	in
                D)      G_DICOMINPUTDIR=$OPTARG         ;;
                E)      Gb_useExpertOptions=1           ;;
                v)      let Gi_verbose=$OPTARG          ;;
		O)      Gb_useOverrideOut=1
			G_OUTDIR=$OPTARG                ;;
		o)      Gb_useOverrideOut=1
                        G_OUTRUNDIR="$OPTARG"           ;;
                p)      G_OUTPREFIX="$OPTARG"           ;;
                R)	G_DIRSUFFIX=$OPTARG		;;
                t)      G_STAGES="$OPTARG"              ;;
		\?)     synopsis_show 
                        exit 0;;
	esac
done

verbosity_check
topDir=$(pwd)

cprint  "hostname"      "[ $(hostname) ]"

## Check on script preconditions
REQUIREDFILES="matlab"
for file in $REQUIREDFILES ; do
        printf "%40s"   "Checking for $file"
        file_checkOnPath $file || fatal fileCheck
done

## Check on input directory and files
statusPrint     "Checking -D <dicomInputDir>"
if [[ "$G_DICOMINPUTDIR" == "-x" ]] ; then fatal noDicomDirArg ; fi
ret_check $?
statusPrint     "Checking on <dicomInputDir>"
dirExist_check $G_DICOMINPUTDIR || fatal noDicomDir
cd $G_DICOMINPUTDIR >/dev/null
G_DICOMINPUTDIR=$(pwd)
cd $topDir

cd ${G_DICOMINPUTDIR}
statusPrint     "Scanning <dicomInputDir>"
ret_check $?
SCANTABLE=$(dcm_mkIndx.bash -t '_' 2>/dev/null)
cd $topDir

if [[ $G_OUTPREFIX == "-x" ]] ; then
  G_OUTPREFIX="$INPUTSCAN-To-$REFSCAN"
fi

statusPrint     "Checking on <outputRunDir>"
if (( !Gb_useOverrideOut )) ; then
  MRID=$(echo "$SCANTABLE" | grep ID | awk '{print $3}')
  G_OUTRUNDIR=${MRID}${G_DIRSUFFIX}/anonymized
fi
dirExist_check ${G_OUTDIR}/$G_OUTRUNDIR "not found - creating"  \
              || mkdir -p ${G_OUTDIR}/$G_OUTRUNDIR              \
              || fatal noOutRunDir
cd ${G_OUTDIR}/$G_OUTRUNDIR >/dev/null
OUTDIR=$(pwd)
lprint "Anonymized directory" 
rprint "[ $OUTDIR ]"
cd $topDir

lprint	"Are we on ubuntu?"
let b_ubuntu=$(grep Ubuntu /etc/issue | wc -l)
if (( b_ubuntu )) ; then
    G_UBUNTUJAVA="MATLAB_JAVA=/usr/lib/jvm/java-6-sun/jre"
fi
rprint	"[ $b_ubuntu ]"

## Check which stages to process
statusPrint     "Checking which stages to process"
barr_stage=([0]=0 [1]=0 [2]=0 [3]=0 [4]=0 [5]=0)
for i in $(seq 1 1) ; do
        b_test=$(expr index $G_STAGES "$i")
        if (( b_test )) ; then b_flag="1" ; else b_flag="0" ; fi
        barr_stage[$i]=$b_flag
done
ret_check $?

G_LOGDIR=$OUTDIR
STAMPLOG=${G_LOGDIR}/${G_SELF}.log
stage_stamp "Init | ($(pwd)) $G_SELF $*" $STAMPLOG

STAGENUM="1"
STAGEPROC=matlab
STAGE=${STAGENUM}-${STAGEPROC}
STAGE1RELDIR=${G_OUTRUNDIR}
STAGE1FULLDIR=${OUTDIR}
statusPrint     "Checking stage 1 output dir"
dirExist_check ${STAGE1FULLDIR} "not found - creating"        \
            || mkdir -p ${STAGE1FULLDIR}                      \
            || fatal noOutRunDir
if (( ${barr_stage[1]} )) ; then
    cd $STAGE1FULLDIR
    statusPrint "$(date) | Processing STAGE 1 - anonymizing DICOM dir | START" "\n"
    statusPrint         "creating MatLAB script file..." "\n"
    MATLABSCRIPT=dcm_anonymize_drive.m
    matlab_scriptCreate $MATLABSCRIPT
    statusPrint         "running MatLAB script file..." "\n"
    
    STAGECMD="eval \"$G_MATLAB -nodesktop -nosplash \
              -r \\\"c = $(basename $MATLABSCRIPT .m)(); exit\\\"\""                                      
    stage_run "$STAGE"  "$STAGECMD"                             \
              "${STAGE1FULLDIR}/${STAGEPROC}.std"               \
              "${STAGE1FULLDIR}/${STAGEPROC}.err"               \
              "SILENT"                                          \
              || fatal stageRun

    statusPrint "$(date) | Processing STAGE 1 - anonymizing DICOM dir | END" "\n"
fi

cd $topDir
verbosity_check
shut_down 0