Compressed CBCT Data

manjula · November 28, 2019, 11:22am

I was today using CBCT data exported as dcm multiframe images of patients.
I used 3D Slicer to import the .dcm file and when it was importing it gave a warning in advance mode saying " Multi-frame image. If slice orientation or spaicing is non-uniform then the image may be displayed incorretly. Use with caution"
And after importing anyway when i use volume rendering to visualize the bones, i realized the rendered image is scaled or compressed or skewed i dont know what exactly but “not proper”.

So i went back and used gdcmconv to uncompressed the .dcm file and used the uncompressed dcm file in the 3D slicer. Then the volume rendering was accurate.

Why is this ? Is there a way of properly importing the compressed cbct data directly to 3D Slicer without using gdcmconv and cli ?

I cannot share the screenshots due to i did not seek permission from the patient to use the data in public.

Thanks

lassoan · November 28, 2019, 12:31pm

This information is not enough to even guess what could have gone wrong. Follow instructions in the DICOM FAQ and if it doesn’t solve the problems then let us know.

manjula · November 28, 2019, 12:34pm

Thank you for the quick reply Prof Iassoan. I will go through it and let you know and come back with more details.

And just now i figure out there is a x3 time scaling added when i uncompressed the dcm with gdcmconv command when i measured the distances.

Is there a way to scale it back or something is horrible wrong from the onset ?

manjula · November 28, 2019, 12:50pm

Thank you Prof @lassoan.

The DICOM Patcher module corrected the problem including the scaling problem i had.

Also just out of interest i tried the volume --> spacing and when i gave same spacing value again i think the problem settled because image spacing in volume information section was 0.3, 0.3 and 1.0 mm so when i changed everything to 0.3 mm it again seems it was working.

Thank you

pieper · November 28, 2019, 4:12pm

@manjula - I’m glad you were very attentive to the spacing issues in the data and were able to find a workable solution. To help make this process more robust, please scan a phantom with known dimensions using the scanner in question and share the data for testing.

lassoan · November 28, 2019, 5:12pm

I guess it’s a Dolphin CBCT scanner. Developers of this device thought that slice thickness tag could be used to define image geometry, which is of course incorrect.

This is the rule in DICOM patcher that fixes it:

github.com

Slicer/Slicer/blob/324ad88c817cc77186a1736a2846a48b06e7a9ae/Modules/Scripted/DICOMPatcher/DICOMPatcher.py#L308


import pydicom


if not hasattr(ds,'NumberOfFrames'):
  return
numberOfFrames = ds.NumberOfFrames
if numberOfFrames <= 1:
  return


# Multi-frame sequence, we may need to add slice positions


# Error in Dolphin 3D CBCT scanners, they store multiple frames but they keep using CTImageStorage as storage class
if ds.SOPClassUID == '1.2.840.10008.5.1.4.1.1.2': # Computed Tomography Image IOD
  ds.SOPClassUID = '1.2.840.10008.5.1.4.1.1.2.1' # Enhanced CT Image IOD


sliceStartPosition = ds.ImagePositionPatient if hasattr(ds,'ImagePositionPatient') else [0,0,0]
sliceAxes = ds.ImageOrientationPatient if hasattr(ds,'ImagePositionPatient') else [1,0,0,0,1,0]
x = sliceAxes[:3]
y = sliceAxes[3:]
z = [x[1] * y[2] - x[2] * y[1], x[2] * y[0] - x[0] * y[2], x[0] * y[1] - x[1] * y[0]] # cross(x,y)
sliceSpacing = ds.SliceThickness if hasattr(ds,'SliceThickness') else 1.0
pixelSpacing = ds.PixelSpacing if hasattr(ds,'PixelSpacing') else [1.0, 1.0]

manjula · November 28, 2019, 5:54pm

@pieper and @lassoan ,

It is a i-CAT KAVO dental CBCT. I will try to scan a phantom with that but it is not my scanner. It belong to colleague of mine and he send me this for a test run with 3D Slicer for a project we are planning for. I am not sure but i will try.

thanks

pieper · November 28, 2019, 6:24pm

Sounds good. Perhaps you can explain to your colleague that the scanner needs to be calibrated/tested to ensure accurate measurements. In any case best of luck.

lassoan · November 28, 2019, 6:55pm

Does the image come directly from the scanner or it has been processed by some software? Maybe the problem is not in the Dolphin software (in previous sample images that I got, Dolphin was in the manufacturer tag, but maybe it was just used for processing).

If it is straight from the scanner then the scanner manufacturer should be made aware of their probable non-conformance with the DICOM standard.

If you can provide a sample scan of a phantom then we can give you more specific information about which required DICOM tags are missing from their images.

manjula · November 29, 2019, 8:30am

I will tell him next week when i meet him.

Is the metdata is of any use ? because It is already available !

lassoan · November 29, 2019, 1:53pm

Yes, metadata is useful, could you copy-paste it here? (make sure you don’t remove any tags just replace values that contain personal information by ***).

manjula · November 29, 2019, 7:47pm

I have pasted two metdata sets.

Metdata Set 01

[0008,0005]	SpecificCharacterSet	ISO_IR 100	CS	10
[0008,0008]	ImageType	[3] ORIGINAL, PRIMARY, AXIAL	CS	22
[0008,0012]	InstanceCreationDate	20170629	DA	8
[0008,0013]	InstanceCreationTime	082742	TM	6
[0008,0014]	InstanceCreatorUID	99999	UI	6
[0008,0016]	SOPClassUID	1.2.840.10008.5.1.4.1.1.2	UI	26
[0008,0018]	SOPInstanceUID	2.16.840.114421.80462.9552036653.9615108653	UI	44
[0008,0020]	StudyDate	20170629	DA	8
[0008,0021]	SeriesDate	20170629	DA	8
[0008,0022]	AcquisitionDate	20170629	DA	8
[0008,0023]	ContentDate	20170629	DA	8
[0008,0030]	StudyTime	082742	TM	6
[0008,0031]	SeriesTime	082742	TM	6
[0008,0032]	AcquisitionTime	082742	TM	6
[0008,0033]	ContentTime	082742	TM	6
[0008,0050]	AccessionNumber		SH	0
[0008,0060]	Modality	CT	CS	2
[0008,0070]	Manufacturer	Imaging Sciences International	LO	30
[0008,0080]	InstitutionName	Imaging Sciences International	LO	30
[0008,0081]	InstitutionAddress	Hatfield, PA	ST	12
[0008,0090]	ReferringPhysicianName		PN	0
[0008,1010]	StationName	RONTGEN	SH	8
[0008,103e]	SeriesDescription		LO	0
[0008,1050]	PerformingPhysicianName		PN	0
[0008,1070]	OperatorsName	rtg	PN	4
[0008,1090]	ManufacturerModelName	17-19	LO	6
[0009,0100]	Unknown Tag & Data	9267	LO	4
[0009,0101]	Unknown Tag & Data	1	IS	2
[0010,0010]	PatientName	*************	PN	16
[0010,0020]	PatientID	**********	LO	10
[0010,0030]	PatientBirthDate	********	DA	8
[0010,0040]	PatientSex	*	CS	2
[0010,2160]	EthnicGroup		SH	0
[0010,4000]	PatientComments		LT	0
[0018,0022]	ScanOptions	PORTRAIT	CS	8
[0018,0050]	SliceThickness	0.300	DS	6
[0018,0060]	KVP	120	DS	4
[0018,0090]	DataCollectionDiameter	170.00	DS	6
[0018,1000]	DeviceSerialNumber	ICU-080462	LO	10
[0018,1020]	SoftwareVersions	1.9.3.13	LO	8
[0018,1150]	ExposureTime	4	IS	2
[0018,1151]	XRayTubeCurrent	5	IS	2
[0018,5100]	PatientPosition		CS	0
[0019,1000]	Unknown Tag & Data	ISI_ACQU_1	LO	10
[0019,1043]	Unknown Tag & Data	1	IS	2
[0019,1050]	Unknown Tag & Data	64.3059095263	DS	14
[0019,1051]	Unknown Tag & Data	73.0000000000	DS	14
[0019,2000]	Unknown Tag & Data	Pan	EFOV	DAP
[0019,2001]	Unknown Tag & Data	9deSGSZl+9BZr9jdo0hvGbDe+I4	LO	28
[0019,2002]	Unknown Tag & Data	4.586000^0.138000^0.138000	ST	50
[0020,000d]	StudyInstanceUID	2.16.840.114421.80462.9552036462	UI	32
[0020,000e]	SeriesInstanceUID	2.16.840.114421.80462.9552036477.9583572477	UI	44
[0020,0010]	StudyID		SH	0
[0020,0011]	SeriesNumber	0	IS	2
[0020,0012]	AcquisitionNumber	0	IS	2
[0020,0013]	InstanceNumber	1	IS	2
[0020,0032]	ImagePositionPatient	[3] -384.000, -384.000, 100.000	DS	26
[0020,0037]	ImageOrientationPatient	[6] 1.000000, 0.000000, 0.000000, 0.000000, 1.000000, 0.000000	DS	54
[0020,0052]	FrameOfReferenceUID	2.16.840.114421.80462.9552036477.9583572477	UI	44
[0020,1040]	PositionReferenceIndicator		LO	0
[0020,1041]	SliceLocation	100.000	DS	8
[0021,1032]	Unknown Tag & Data	18.537371	DS	10
[0021,1034]	Unknown Tag & Data	8.883619	DS	8
[0028,0002]	SamplesPerPixel	1	US	2
[0028,0004]	PhotometricInterpretation	MONOCHROME2	CS	12
[0028,0008]	NumberOfFrames	576	IS	4
[0028,0010]	Rows	768	US	2
[0028,0011]	Columns	768	US	2
[0028,0030]	PixelSpacing	[2] 0.300, 0.300	DS	12
[0028,0100]	BitsAllocated	16	US	2
[0028,0101]	BitsStored	16	US	2
[0028,0102]	HighBit	15	US	2
[0028,0103]	PixelRepresentation	1	US	2
[0028,1050]	WindowCenter	1100	DS	4
[0028,1051]	WindowWidth	4500	DS	4
[0028,1052]	RescaleIntercept	0	DS	2
[0028,1053]	RescaleSlope	1	DS	2
[0032,4000]	RETIRED_StudyComments		LT	0
[7fe0,0010]	PixelData		OB	0

MetData set 02

[0008,0005]	SpecificCharacterSet	ISO_IR 100	CS	10
[0008,0008]	ImageType	[3] ORIGINAL, PRIMARY, AXIAL	CS	22
[0008,0012]	InstanceCreationDate	20170914	DA	8
[0008,0013]	InstanceCreationTime	092126	TM	6
[0008,0014]	InstanceCreatorUID	99999	UI	6
[0008,0016]	SOPClassUID	1.2.840.10008.5.1.4.1.1.2	UI	26
[0008,0018]	SOPInstanceUID	2.16.840.114421.80462.9558692675.9621764675	UI	44
[0008,0020]	StudyDate	20170914	DA	8
[0008,0021]	SeriesDate	20170914	DA	8
[0008,0022]	AcquisitionDate	20170914	DA	8
[0008,0023]	ContentDate	20170914	DA	8
[0008,0030]	StudyTime	092126	TM	6
[0008,0031]	SeriesTime	092126	TM	6
[0008,0032]	AcquisitionTime	092126	TM	6
[0008,0033]	ContentTime	092126	TM	6
[0008,0050]	AccessionNumber		SH	0
[0008,0060]	Modality	CT	CS	2
[0008,0070]	Manufacturer	Imaging Sciences International	LO	30
[0008,0080]	InstitutionName	Imaging Sciences International	LO	30
[0008,0081]	InstitutionAddress	Hatfield, PA	ST	12
[0008,0090]	ReferringPhysicianName		PN	0
[0008,1010]	StationName	RONTGEN	SH	8
[0008,103e]	SeriesDescription		LO	0
[0008,1050]	PerformingPhysicianName		PN	0
[0008,1070]	OperatorsName	rtg	PN	4
[0008,1090]	ManufacturerModelName	17-19	LO	6
[0009,0100]	Unknown Tag & Data	8809	LO	4
[0009,0101]	Unknown Tag & Data	1	IS	2
[0010,0010]	PatientName	********	PN	12
[0010,0020]	PatientID	********	LO	10
[0010,0030]	PatientBirthDate	********	DA	8
[0010,0040]	PatientSex	*	CS	2
[0010,2160]	EthnicGroup		SH	0
[0010,4000]	PatientComments		LT	0
[0018,0022]	ScanOptions	PORTRAIT	CS	8
[0018,0050]	SliceThickness	0.300	DS	6
[0018,0060]	KVP	120	DS	4
[0018,0090]	DataCollectionDiameter	170.00	DS	6
[0018,1000]	DeviceSerialNumber	ICU-080462	LO	10
[0018,1020]	SoftwareVersions	1.9.3.13	LO	8
[0018,1150]	ExposureTime	4	IS	2
[0018,1151]	XRayTubeCurrent	5	IS	2
[0018,5100]	PatientPosition		CS	0
[0019,1000]	Unknown Tag & Data	ISI_ACQU_1	LO	10
[0019,1043]	Unknown Tag & Data	1	IS	2
[0019,1050]	Unknown Tag & Data	77.6498551190	DS	14
[0019,1051]	Unknown Tag & Data	124.0000000000	DS	14
[0019,2000]	Unknown Tag & Data	Pan	EFOV	DAP
[0019,2001]	Unknown Tag & Data	o6fVI9wQoGd8/4ckK5VByBD58Sg	LO	28
[0019,2002]	Unknown Tag & Data	4.586000^0.138000^0.138000^0.138000	ST	58
[0020,000d]	StudyInstanceUID	2.16.840.114421.80462.9558692486	UI	32
[0020,000e]	SeriesInstanceUID	2.16.840.114421.80462.9558692501.9590228501	UI	44
[0020,0010]	StudyID		SH	0
[0020,0011]	SeriesNumber	0	IS	2
[0020,0012]	AcquisitionNumber	0	IS	2
[0020,0013]	InstanceNumber	1	IS	2
[0020,0032]	ImagePositionPatient	[3] -384.000, -384.000, 100.000	DS	26
[0020,0037]	ImageOrientationPatient	[6] 1.000000, 0.000000, 0.000000, 0.000000, 1.000000, 0.000000	DS	54
[0020,0052]	FrameOfReferenceUID	2.16.840.114421.80462.9558692501.9590228501	UI	44
[0020,1040]	PositionReferenceIndicator		LO	0
[0020,1041]	SliceLocation	100.000	DS	8
[0021,1032]	Unknown Tag & Data	18.537371	DS	10
[0021,1034]	Unknown Tag & Data	8.883619	DS	8
[0028,0002]	SamplesPerPixel	1	US	2
[0028,0004]	PhotometricInterpretation	MONOCHROME2	CS	12
[0028,0008]	NumberOfFrames	576	IS	4
[0028,0010]	Rows	768	US	2
[0028,0011]	Columns	768	US	2
[0028,0030]	PixelSpacing	[2] 0.300, 0.300	DS	12
[0028,0100]	BitsAllocated	16	US	2
[0028,0101]	BitsStored	16	US	2
[0028,0102]	HighBit	15	US	2
[0028,0103]	PixelRepresentation	1	US	2
[0028,1050]	WindowCenter	1100	DS	4
[0028,1051]	WindowWidth	4500	DS	4
[0028,1052]	RescaleIntercept	0	DS	2
[0028,1053]	RescaleSlope	1	DS	2
[0032,4000]	RETIRED_StudyComments		LT	0
[7fe0,0010]	PixelData		OB	0

manjula · December 2, 2019, 8:35pm

@lassoan @pieper

Jsut out of curiosity, did you manage to obtain any useful information from the metdata ? or do I have to see the possibility of a phantom object scan ?

lassoan · December 3, 2019, 4:27am

Sorry for the slow response, both @pieper and I are quite busy at RSNA. I’ve checked the metadata and the problem is the following:

Slice position must be specified for each slice. However in the ISI i-CAT device, only one position is specified using [0020,0032] ImagePositionPatient and probably the developers thought that by adding a [0018,0050] SliceThickness tag will specify slice positions within the image. That is incorrect. Instead, values that are the same for all frames (image orientation, pixel spacing, etc.) should be specified in tag (5200,9229) and values that are different in each slice (such as fields position) should be specified in tag (5200,9230). For example:

(5200,9229) SQ (Sequence with undefined length #=1)     # u/l, 1 SharedFunctionalGroupsSequence
(0020,9116) SQ (Sequence with undefined length #=1)     # u/l, 1 PlaneOrientationSequence
    (0020,0037) DS [1.00000\0.00000\0.00000\0.00000\1.00000\0.00000] #  48, 6 ImageOrientationPatient
(0028,9110) SQ (Sequence with undefined length #=1)     # u/l, 1 PixelMeasuresSequence
    (0018,0050) DS [3.00000]                                #   8, 1 SliceThickness
    (0028,0030) DS [0.597656\0.597656]                      #  18, 2 PixelSpacing
...

(5200,9230) SQ (Sequence with undefined length #=54)    # u/l, 1 PerFrameFunctionalGroupsSequence
(0020,9113) SQ (Sequence with undefined length #=1)     # u/l, 1 PlanePositionSequence
    (0020,0032) DS [-94.7012\-312.701\-806.500]             #  26, 3 ImagePositionPatient
(0020,9113) SQ (Sequence with undefined length #=1)     # u/l, 1 PlanePositionSequence
    (0020,0032) DS [-94.7012\-312.701\-809.500]             #  26, 3 ImagePositionPatient
...

manjula · December 3, 2019, 1:17pm

Thank you for the information. It has been very useful.

Further, I have noticed that few CBCT’s that i have worked with in the past to import data in to 3D Slicer is having the same problem but did not knew what was wrong at the time but never with the conventional CT’s.

Are CBCT manufacturers are notoriously not compiling to DICOM format ? is this something non compliance is mainly a problem with CBCT manufacturers ?

lassoan · December 3, 2019, 1:44pm

Yes, CBCT and microCT manufacturers often have DICOM compliance issues. Maybe they have fixed the problems in their current products but not released updates for the earlier models.

Topic		Replies	Views
3D volume appears distorted - jaw looks elongated Support dicom	3	705	November 12, 2019
Import CBCT image volume from DICOM Support dicom	1	526	August 24, 2018
CT Image Scan Error Support	3	566	May 31, 2021
DICOM cardiac CT image time series appears as moving line Support dicom , heart	4	1131	February 25, 2020
Annotation Ruler Support	14	2604	July 21, 2018

Compressed CBCT Data

Related topics