dicom conversion script

README.md

@@ -16,6 +16,11 @@
 
 ## Image pipeline
 
+### Preprocessing
+
+originally DICOMs were converted to 299x299 PNGs using `convert_dicom_list_to_png.sh` script
+
+
 ### Weight files are available [here](https://datashare.ucsf.edu/stash/dataset/doi:10.7272/Q6XK8CQ9)
 
 ### General image model

convert_dicom_list_to_png.sh

@@ -0,0 +1,20 @@
+#!/bin/bash
+filelist=$1
+OUT_DIRECTORY="/data/UCSF_MAMMO/2017-07-png-16bit"
+TMPFILE="tmp.dcm"
+
+nn=0
+for file in $filelist; do
+    nn=$((nn+1))
+    dirn_=$(dirname $file)
+    id=`echo $file | awk -v FS="/" '{for (nn=NF-3;nn<NF;nn++){ printf $nn "_" }; print $NF }'`
+    id=${id%.dcm}
+    #echo  $id; 
+    # gdcmconv --raw "$file" $TMPFILE
+    # convert $TMPFILE -resize 299x299! $OUT_DIRECTORY/$id.png
+    gdcmconv --raw "$file" /dev/stdout | convert /dev/stdin -depth 16 -resize 299x299! $OUT_DIRECTORY/$id.png
+    #-depth
+    if [ $((nn % 40)) -eq 0 ]; then
+    echo "$nn"
+fi
+done