dicom_redact.py - added lossless compression

doc/dicom_redact.md

@@ -46,18 +46,20 @@ overlays. This tool can do that job to fully clean a DICOM file.
 ## NOTE!
 
 The utility has to decompress the image (if compressed), but it
-does not (yet) recompress the image again afterwards, so the file size
-may increase. Some other redaction tools have the ability to preserve
+does not recompress the image again afterwards using the same
+compression scheme so the file size may increase.
+Some other redaction tools have the ability to preserve
 most of a lossy-JPEG-compressed image except for the redacted blocks,
-but those tools do not handle overlays at all. Compression, and lossy
-compression support can of course be added to this tool later.
+but those tools do not handle overlays at all. Lossless compression is
+an option but the only scheme supported is JPEG2000Lossless.
 
 ## Usage:
 
 ```
   -v, --verbose         Verbose
   --db DB               database directory to read rectangles (needs --dicom)
   --csv CSV             CSV path to read rectangles (redacts all files in csv if --dicom not used)
+  --compress            Compress the output using lossless JPEG2000
   --dicom DICOM         DICOM path to be redacted
   -o OUTPUT, --output O Output DICOM dir or filename (created automatically if not specified)
   --relative-path RELATIVE

src/applications/dicom_redact.py

@@ -20,6 +20,7 @@
 import re
 import pydicom
 from pydicom.pixel_data_handlers.numpy_handler import pack_bits
+from pydicom.uid import JPEGLSLossless, JPEG2000Lossless
 from DicomPixelAnon.rect import Rect, DicomRect, DicomRectText, rect_exclusive_list
 from DicomPixelAnon.nerengine import NER
 from DicomPixelAnon.ocrenum import OCREnum
@@ -54,6 +55,10 @@
 elem_OverlayNumFrames = 0x0015
 elem_OverlayOrigin = 0x0050
 
+# Choose compression scheme as 'JPEG-LS' or 'JPEG2000'
+# (only the latter is supported on Python 3.6)
+compression_scheme = 'JPEG2000'
+
 
 # ---------------------------------------------------------------------
 
@@ -71,6 +76,21 @@ def mark_as_uncompressed(ds):
         ds.file_meta.TransferSyntaxUID = pydicom.uid.ExplicitVRBigEndian
 
 
+# ---------------------------------------------------------------------
+
+def compress_dataset(ds):
+    """ Do lossless compression of the pydicom Dataset
+    """
+    if compression_scheme == 'JPEG-LS':
+        ds.compress(JPEGLSLossless)
+        # ds.file_meta.TransferSyntaxUID = JPEGLSLossless # should already be done
+    elif compression_scheme == 'JPEG2000':
+        ds.compress(JPEG2000Lossless)
+        # ds.file_meta.TransferSyntaxUID = JPEG2000Lossless # should already be done
+    else:
+        logger.warning('unknown compression_scheme, will not compress')
+
+
 # ---------------------------------------------------------------------
 
 def overlay_tag_group_from_index(overlay):
@@ -647,6 +667,7 @@ def test_create_output_filename():
     parser = argparse.ArgumentParser(description='Redact DICOM')
     parser.add_argument('-v', '--verbose', action="store_true", help='Verbose')
     parser.add_argument('--db', dest='db', action="store", help='database directory to read rectangles (needs --dicom)')
+    parser.add_argument('--compress', dest='compress', action="store_true", help='Use lossless compression (JPEG2000)')
     parser.add_argument('--csv', dest='csv', action="store", help='CSV path to read rectangles (redacts all files in csv if --dicom not used)')
     parser.add_argument('--dicom', dest='dicom', nargs='*', action="store", help='DICOM filename(s) to be redacted', default=None)
     parser.add_argument('-o', '--output', dest='output', action="store", help='Output DICOM dir or filename (created automatically if not specified)', default=[])
@@ -724,4 +745,6 @@ def test_create_output_filename():
         outfilename = create_output_filename(infilename, args.output, args.relative)
         ds = pydicom.dcmread(infilename)
         redact_DicomRect_rectangles(ds, rect_list)
+        if args.compress:
+            compress_dataset(ds)
         ds.save_as(outfilename)

src/requirements.txt

@@ -1,6 +1,7 @@
 pydicom
 pylibjpeg
 pylibjpeg_libjpeg
+pylibjpeg_openjpeg
 pydal
 pymongo
 Pillow