Dear experts,
I got a problem with slice-timing data extraction using heudiconv on Philips DICOMs. Indeed, when passing directory to BIDS validator, an error message told me that slice-timing information were missing.
My RS-fMRI DICOMs data have dim3=10000 and dim4=1 based on dicomsinfo.tsv, and when converting to nifti dim4=250. Could this error be based on the particular format of initial DICOM files ?
Thanks !
Can you share the command you are using to run heudiconv, your dicominfo.tsv, and the heuristic you are using?
Hi @mgxd,
Please find below the command I use:
docker run --rm -it -v /mnt/HDD_2To/COMAJ/BIDS/DICOM:/data:ro \
-v /mnt/HDD_2To/COMAJ/BIDS/Nifti:/output nipy/heudiconv:latest \
-d /data/{subject}/*/*.dcm -s 207187_M0_2016-10-11 \
-f /data/BIDS_heuristic_COMAJ.py -b -o /output
Below my dicominfo.tsv:
| total_files_till_now | example_dcm_file | series_id | dcm_dir_name | unspecified2 | unspecified3 | dim1 | dim2 | dim3 | dim4 | TR | TE | protocol_name | is_motion_corrected | is_derived | patient_id | study_description | referring_physician_name | series_description | sequence_name | image_type | accession_number | patient_age | patient_sex | date | series_uid |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 11 | MR_1.3.46.670589.11.34037.5.0.4548.2016101113143433305.dcm | 101-WIP SURVEY CLEAR | 101_SURVEY | - | - | 256 | 256 | 11 | 1 | 0.0044959002 | 2.303 | WIP SURVEY CLEAR | False | False | 3T11101613H15 | COMAJ | SURVEY | Not found | (‘ORIGINAL’, ‘PRIMARY’, ‘M_FFE’, ‘M’, ‘FFE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.4548.2016101113141954303 | |
| 3851 | MR_1.3.46.670589.11.34037.5.0.4548.2016101113252254320.dcm | 201-WIP 3DMULTIGRE SAG STRICT SENSE | 201_3DMULTIGRE_SAG_STRICT | - | - | 256 | 256 | 3840 | 1 | 0.054 | 30.521 | WIP 3DMULTIGRE SAG STRICT SENSE | False | False | 3T11101613H15 | COMAJ | 3DMULTIGRE SAG STRICT | Not found | (‘ORIGINAL’, ‘PRIMARY’, ‘PHASE MAP’, ‘P’, ‘FFE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.4548.2016101113151209318 | |
| 4011 | MR_1.3.46.670589.11.34037.5.0.4548.2016101113370035006.dcm | 301-WIP s3DT1 ISO 1mm HR SENSE | 301_s3DT1_ISO_1mm_HR | - | - | 256 | 256 | 160 | 1 | 0.0098773003 | 4.595 | WIP s3DT1 ISO 1mm HR SENSE | False | False | 3T11101613H15 | COMAJ | s3DT1 ISO 1mm HR | Not found | (‘ORIGINAL’, ‘PRIMARY’, ‘M_FFE’, ‘M’, ‘FFE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.4548.2016101113252474529 | |
| 4034 | MR_1.3.46.670589.11.34037.5.0.5740.2016101113370804000.dcm | 302-ax | 302_ax | - | - | 256 | 256 | 23 | 1 | 0.0098773003 | 4.595 | ax | False | False | 3T11101613H15 | COMAJ | ax | (‘ORIGINAL’, ‘PRIMARY’, ‘PROJECTION IMAGE’, ‘M’, ‘FFE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.2580.2016101113370746000 | ||
| 4057 | MR_1.3.46.670589.11.34037.5.0.7788.2016101113371042000.dcm | 303-coro | 303_coro | - | - | 256 | 256 | 23 | 1 | 0.0098773003 | 4.595 | coro | False | False | 3T11101613H15 | COMAJ | coro | (‘ORIGINAL’, ‘PRIMARY’, ‘PROJECTION IMAGE’, ‘M’, ‘FFE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.9836.2016101113370988000 | ||
| 4080 | MR_1.3.46.670589.11.34037.5.0.9376.2016101113371261000.dcm | 304-sag | 304_sag | - | - | 256 | 256 | 23 | 1 | 0.0098773003 | 4.595 | sag | False | False | 3T11101613H15 | COMAJ | sag | (‘ORIGINAL’, ‘PRIMARY’, ‘PROJECTION IMAGE’, ‘M’, ‘FFE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.8664.2016101113371205000 | ||
| 4111 | MR_1.3.46.670589.11.34037.5.0.4548.2016101113402658168.dcm | 401-WIP T2 * CLEAR | 401_T2_ETOILE | - | - | 256 | 256 | 31 | 1 | 0.9140009155 | 16.12 | WIP T2 * CLEAR | False | False | 3T11101613H15 | COMAJ | T2 * | Not found | (‘ORIGINAL’, ‘PRIMARY’, ‘M_FFE’, ‘M’, ‘FFE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.4548.2016101113370239049 | |
| 4141 | MR_1.3.46.670589.11.34037.5.0.4548.2016101113412429201.dcm | 501-WIP TSE T2 5mm CLEAR | 501_TSE_T2_5mm | - | - | 560 | 560 | 30 | 1 | 2.8172145996 | 80 | WIP TSE T2 5mm CLEAR | False | False | 3T11101613H15 | COMAJ | TSE T2 5mm | Not found | (‘ORIGINAL’, ‘PRIMARY’, ‘M_SE’, ‘M’, ‘SE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.4548.2016101113402920199 | |
| 4185 | MR_1.3.46.670589.11.34037.5.0.4548.2016101113441543233.dcm | 601-WIP FLAIR_3 mm SENSE | 601_FLAIR_3_mm | - | - | 560 | 560 | 44 | 1 | 11 | 125 | WIP FLAIR_3 mm SENSE | False | False | 3T11101613H15 | COMAJ | FLAIR_3 mm | Not found | (‘ORIGINAL’, ‘PRIMARY’, ‘M_IR’, ‘M’, ‘IR’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.4548.2016101113421765231 | |
| 5145 | MR_1.3.46.670589.11.34037.5.0.4548.2016101113543018279.dcm | 701-WIP DTI_15dir serie 1 SENSE | 701_DTI_15dir_serie_1 | - | - | 128 | 128 | 960 | 1 | 12 | 56 | WIP DTI_15dir serie 1 SENSE | False | False | 3T11101613H15 | COMAJ | DTI_15dir serie 1 | Not found | (‘ORIGINAL’, ‘PRIMARY’, ‘M_SE’, ‘M’, ‘SE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.4548.2016101113492972277 | |
| 5205 | MR_1.3.46.670589.11.34037.5.0.6520.2016101115153538035.dcm | 702-isoTRACE | 702_isoTRACE | - | - | 128 | 128 | 60 | 1 | 12 | 56 | isoTRACE | False | False | 3T11101613H15 | COMAJ | isoTRACE | (‘ORIGINAL’, ‘PRIMARY’, ‘M_SE’, ‘M’, ‘SE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.6520.2016101115153521000 | ||
| 5685 | MR_1.3.46.670589.11.34037.5.0.4548.2016101113570730241.dcm | 801-WIP DTI CORRECTION SENSE | 801_DTI_CORRECTION | - | - | 128 | 128 | 480 | 1 | 12 | 56 | WIP DTI CORRECTION SENSE | False | False | 3T11101613H15 | COMAJ | DTI CORRECTION | Not found | (‘ORIGINAL’, ‘PRIMARY’, ‘M_SE’, ‘M’, ‘SE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.4548.2016101113543280731 | |
| 6645 | MR_1.3.46.670589.11.34037.5.0.4548.2016101114013884723.dcm | 901-WIP DTI_15dir serie 2 SENSE | 901_DTI_15dir_serie_2 | - | - | 128 | 128 | 960 | 1 | 12 | 56 | WIP DTI_15dir serie 2 SENSE | False | False | 3T11101613H15 | COMAJ | DTI_15dir serie 2 | Not found | (‘ORIGINAL’, ‘PRIMARY’, ‘M_SE’, ‘M’, ‘SE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.4548.2016101113570999626 | |
| 6705 | MR_1.3.46.670589.11.34037.5.0.4548.2016101114070249687.dcm | 1001-WIP B0 map CLEAR | 1001_B0_map | - | - | 128 | 128 | 60 | 1 | 0.5899979858 | 3.947 | WIP B0 map CLEAR | False | False | 3T11101613H15 | COMAJ | B0 map | Not found | (‘ORIGINAL’, ‘PRIMARY’, ‘M_FFE’, ‘M’, ‘FFE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.4548.2016101114014381683 | |
| 16705 | MR_1.3.46.670589.11.34037.5.0.4548.2016101114075631749.dcm | 1101-WIP FE_EPI 64x64 resting state SENSE | 1101_FE_EPI_64x64_resting_state | - | - | 64 | 64 | 10000 | 1 | 2.4 | 30 | WIP FE_EPI 64x64 resting state SENSE | False | False | 3T11101613H15 | COMAJ | FE_EPI 64x64 resting state | Not found | (‘ORIGINAL’, ‘PRIMARY’, ‘M_FFE’, ‘M’, ‘FFE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.4548.2016101114071031747 | |
| 16745 | MR_1.3.46.670589.11.34037.5.0.4548.2016101114190599751.dcm | 1201-WIP SE_FMRI_CORR_PA SENSE | 1201_SE_FMRI_CORR_PA | - | - | 64 | 64 | 40 | 1 | 7.06 | 58 | WIP SE_FMRI_CORR_PA SENSE | False | False | 3T11101613H15 | COMAJ | SE_FMRI_CORR_PA | Not found | (‘ORIGINAL’, ‘PRIMARY’, ‘M_SE’, ‘M’, ‘SE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.4548.2016101114175419749 | |
| 16785 | MR_1.3.46.670589.11.34037.5.0.4548.2016101114201142793.dcm | 1301-WIP SE_FMRI_CORR_AP SENSE | 1301_SE_FMRI_CORR_AP | - | - | 64 | 64 | 40 | 1 | 7.06 | 58 | WIP SE_FMRI_CORR_AP SENSE | False | False | 3T11101613H15 | COMAJ | SE_FMRI_CORR_AP | Not found | (‘ORIGINAL’, ‘PRIMARY’, ‘M_SE’, ‘M’, ‘SE’) | 3T11101613H15 | 066Y | F | 20161011 | 1.3.46.670589.11.34037.5.0.4548.2016101114190634791 |
And finally the heuristic:
import os
def create_key(template, outtype=('nii.gz',), annotation_classes=None):
if template is None or not template:
raise ValueError('Template must be a valid format string')
return (template, outtype, annotation_classes)
def infotodict(seqinfo):
"""Heuristic evaluator for determining which runs belong where
allowed template fields - follow python string module:
item: index within category
subject: participant id
seqitem: run number during scanning
subindex: sub index within group
"""
t1 = create_key('sub-{subject}/anat/sub-{subject}_T1w')
t2 = create_key('sub-{subject}/anat/sub-{subject}_T2w')
rest = create_key('sub-{subject}/func/sub-{subject}_task-rest_run-{item:02d}_bold')
rest_corrpa = create_key('sub-{subject}/func/sub-{subject}_task-rest_run-{item:02d}_corrpa')
rest_corrap = create_key('sub-{subject}/func/sub-{subject}_task-rest_run-{item:02d}_corrap')
dwi = create_key('sub-{subject}/dwi/sub-{subject}_run-{item:02d}_dwi')
dwi_corr = create_key('sub-{subject}/dwi/sub-{subject}_run-{item:02d}_corr')
fmap = create_key('sub-{subject}/fmap/sub-{subject}_dir-{dir}_run-{item:02d}_epi')
info = {t1:[], t2:[],
rest:[], dwi:[],
rest_corrpa:[], rest_corrap:[], dwi_corr:[],
fmap:[]}
for idx, s in enumerate(seqinfo):
# T1 and T2 scans
if (s.dim3 == 160) and (s.dim4 == 1) and ('s3DT1 ISO 1mm' in s.protocol_name):
info[t1] = [s.series_id]
if (s.dim3 == 44) and (s.dim4 == 1) and ('FLAIR_3 mm SENSE' in s.protocol_name):
info[t2] = [s.series_id]
# diffusion scans
if ('DTI' in s.protocol_name):
key = None
if (s.dim3 == 960):
key = dwi
elif (s.dim3 == 480) and ('DTI CORRECTION' in s.series_description):
key = dwi_corr
if key:
info[key].append({'item': s.series_id})
# functional scans
if (s.dim3 == 10000) and ('EPI 64x64 resting state' in s.protocol_name):
info[rest].append({'item': s.series_id})
if (s.dim3 == 40) and ('FMRI_CORR_PA' in s.protocol_name):
info[rest_corrpa].append({'item': s.series_id})
if (s.dim3 == 40) and ('FMRI_CORR_AP' in s.protocol_name):
info[rest_corrap].append({'item': s.series_id})
if (s.dim3 == 60) and ('B0 map' in s.protocol_name):
dirtype = s.protocol_name.split('_')[-1]
info[fmap].append({'item': s.series_id, 'dir': dirtype})
# You can even put checks in place for your protocol
msg = []
if len(info[t1]) != 1: msg.append('Missing correct number of t1 runs')
if len(info[t2]) != 1: msg.append('Missing correct number of t2 runs')
if len(info[dwi]) != 2: msg.append('Missing correct number of dwi runs')
if len(info[dwi_corr]) != 1: msg.append('Missing correct number of dwi_corr runs')
if len(info[rest]) != 1: msg.append('Missing correct number of resting runs')
if len(info[rest_corrpa]) != 1: msg.append('Missing correct number of rest_corrpa runs')
if len(info[rest_corrap]) != 1: msg.append('Missing correct number of rest_corrap runs')
if msg:
raise ValueError('\n'.join(msg))
return info
Thanks !
Ah, it looks like heudiconv is having a problem accurately extracting information from these Philips scans - there is an open issue on our github repo regarding this.
However, if dcm2niix is able to extract the slicetiming information, your heudiconv conversion should work (in theory).
rest_corrpa = create_key(‘sub-{subject}/func/sub-{subject}_task-rest_run-{item:02d}_corrpa’)
rest_corrap = create_key(‘sub-{subject}/func/sub-{subject}_task-rest_run-{item:02d}_corrap’)
dwi_corr = create_key(‘sub-{subject}/dwi/sub-{subject}_run-{item:02d}_corr’)
I don’t believe these are in valid bids form - instead, you may want to change those to:
rest = create_key('sub-{subject}/func/sub-{subject}_task-rest_run-{item:02d}_bold')
dwi = create_key('sub-{subject}/dwi/sub-{subject}_run-{item:02d}_dwi')
The problem is not with the converter but the source images. Philips DICOM images do not store slice timing information. There are no official DICOM tags to report this information - on Siemens scanners we have reverse engineered the proprietary CSA header. However, Philips scanners do not currently store this information (as confirmed recently by Matthew Clemence from Philips). I encourage users at Philips sites to lobby their research collaboration managers to include this information in future software releases.
In brief, dcm2niix and other converters are only able to convert information that is present in the source images. Until Philips provides this information, we will be unable to automatically populate these fields. Fortunately, these parameters are pretty easy to estimate and consistent across runs of the same sequence. Be aware that if you use prospective motion correction it adds a slight pause at the end of the volume, and so the actual timing may be a tiny bit different from the canonical formulas, but the difference is so small it should have no influence.
Ok @mgxd so heudiconv docker image should work if dcm2niix manages to extract slice-timing ? Else what should I have to do ?
Concerning the non BIDS valid names, I have other posts talking about these supplementary files used to SDC and how to manage them.
Thanks @Chris_Rorden for these details !
Could you provide me links to how fill slice-timing field by estimation of these parameters ?
I can only re-confirm that we had no slice timing information stored anywhere in DICOMs for our previous Philips (achieva?) magnet . The only thing we could deduce is the slice order (based on some of the UIDs) an “fill in” the TR accordingly to get slice timing for the believers 
My web page includes some Matlab scripts to help. It also describes how you can deduce the slice order by acquiring data while moving your head in a known direction.
What you need to know:
Consider the case where the TR is 3 seconds, there are 36 slices, ascending slice order, continuous acquisition without a pause, no multiband. In this case the Matlab code would be
TRsec = 3
nSlices = 36
TA = TRsec/nSlices
bidsSliceTiming=[0:TA:TRsec-TA]
I keep notes on Philips DICOM and PAR/REC conversion. These can help explain why dcm2niix is unable to populate all of the BIDS fields for these images, and provides alternative suggestions.
Thank you for the explanation.
@Chris_Rorden @yarikoptic
Assuming that there is no temporal gap between volumes (sparse imaging, prospective slice time correction), that this is a Philips scanner (as Siemens scanners use different interleaving depending on odd vs even number of slices), and that this is not multi-band, the Matlab script:
TRsec = 3;
nSlices = 48;
TA = TRsec/nSlices; %assumes no temporal gap between volumes
bidsSliceTiming=[0:TA:TRsec-TA]; %ascending
if false %descending
bidsSliceTiming = flip(bidsSliceTiming);
end
if true %interleaved
order = [1:2:nSlices 2:2:nSlices]
bidsSliceTiming(order) = bidsSliceTiming;
end
%report results
fprintf(' "SliceTiming": [\n');
for i = 1 : nSlices
fprintf(' %g', bidsSliceTiming(i));
if (i < nSlices)
fprintf(',\n');
else
fprintf(' ],\n');
end
end
Generates
"SliceTiming": [
0,
1.5,
0.0625,
1.5625,
0.125,
1.625,
0.1875,
1.6875,
0.25,
1.75,
0.3125,
1.8125,
0.375,
1.875,
0.4375,
1.9375,
0.5,
2,
0.5625,
2.0625,
0.625,
2.125,
0.6875,
2.1875,
0.75,
2.25,
0.8125,
2.3125,
0.875,
2.375,
0.9375,
2.4375,
1,
2.5,
1.0625,
2.5625,
1.125,
2.625,
1.1875,
2.6875,
1.25,
2.75,
1.3125,
2.8125,
1.375,
2.875,
1.4375,
2.9375 ],Thank you so much!! I really appreciated.
are you 100% sure that it was interleaved like that and not by “sqrt(nslices)” as it was in our philips?
FWIW – I found that elderly script of mine which tries to deduce the order of the slices in the philips dicom: http://www.onerussian.com/tmp/philips_order.py
fix it up for recent pydicom or use with dicom==0.9.9.post1 . For some sample recent DICOMs from Philips I got
$> python ~/trash/philips_order.py XX_0016
/home/yoh/proj/repronim/umass/dicoms/David2_En/DICOM/venvs-old-dicom/local/lib/python2.7/site-packages/dicom/__init__.py:53: UserWarning:
This code is using an older version of pydicom, which is no longer
maintained as of Jan 2017. You can access the new pydicom features and API
by installing `pydicom` from PyPI.
See 'Transitioning to pydicom 1.x' section at pydicom.readthedocs.org
for more information.
warnings.warn(msg)
XX_0016
Series: func_task-rest_run-01
Sequence: FEEPI
Failed to figure out for XX_0016 (skipped): Dataset does not have attribute 'PerFrameFunctionalGroupsSequence'.
(venvs-old-dicom) 1 27272 [1].....................................:Tue 17 Sep 2019 12:41:03 PM EDT:.
(git-annex)hopa:~/proj/repronim/umass/dicoms/David2_En[master]DICOM
$> python ~/trash/philips_order.py IM_0027
/home/yoh/proj/repronim/umass/dicoms/David2_En/DICOM/venvs-old-dicom/local/lib/python2.7/site-packages/dicom/__init__.py:53: UserWarning:
This code is using an older version of pydicom, which is no longer
maintained as of Jan 2017. You can access the new pydicom features and API
by installing `pydicom` from PyPI.
See 'Transitioning to pydicom 1.x' section at pydicom.readthedocs.org
for more information.
warnings.warn(msg)
IM_0027
Series: Patient Aligned MPR AWPLAN_SMARTPLAN_TYPE_BRAIN
Sequence: T1TFE
# slices: 54
order: [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53]
TR (sec): 0.00
timing???: [0.0e+00 1.0e-05 2.0e-05 3.0e-05 4.0e-05 5.0e-05 6.0e-05 7.0e-05 8.0e-05 9.0e-05 1.0e-04 1.1e-04 1.2e-04 1.3e-04 1.4e-04 1.5e-04 1.6e-04 1.7e-04 1.8e-04 1.9e-04 2.0e-04 2.1e-04 2.2e-04 2.3e-04 2.4e-04 2.5e-04 2.6e-04 2.7e-04 2.8e-04 2.9e-04 3.0e-04 3.1e-04 3.2e-04 3.3e-04 3.4e-04 3.5e-04 3.6e-04 3.7e-04 3.8e-04 3.9e-04 4.0e-04 4.1e-04 4.2e-04 4.3e-04 4.4e-04 4.5e-04 4.6e-04 4.7e-04 4.8e-04 4.9e-04 5.0e-04 5.1e-04 5.2e-04 5.3e-04]
max(???): 0.00
so might need more fixups to work for your EPI
Thank you so much!
I’m working on resting state fMRI (Phillips) from ADNI (http://adni.loni.usc.edu/data-samples/access-data/).
One subject has 6720 .dcm files and the dimension of one fMRI data is [64, 64, 48, 140].
I tried to use your script, however, I had some problems.
The error message said that my data has no “PerFrameFunctionalGroupsSequence”.
When I read dicom file through pydicom, the dicom file needs to be one 4D fMRI data?
PathDicom = "/home/rrt/MyProject/Dicom2/003_S_5019/2013_01_01/fMRI/"
lstFilesDCM = []
for dirName, subdirList, fileList in os.walk(PathDicom):
for filename in fileList:
if ".dcm" in filename.lower(): # check whether the file's DICOM
lstFilesDCM.append(os.path.join(dirName,filename))
ds = pydicom.dcmread(lstFilesDCM[0])
print(ds)
One of this results shows that "(0018, 0023) MR Acquisition Type CS: ‘2D’ ".
What should I do?
Is it possible to read 6720 files at once?
Or do I have to concatenate 6720 files to one 4D .dcm file? For this case, I have no information about the order of 6720 files.
Accordingly, there are no FrameContentSequence, PlanePositionSequence…
well, in your case my wild guess would be that files are in the order of acquisition, so you could look for some position fields (ImagePositionPatient) and see if they differ.
I would have dcmdump’ed two neighboring files and checked which fields differ and how . But may be someone has a better/cooked up complete solution already? sorry - I cannot dig into it ATM myself
I just looked at this method using Philips data from Baxter Rogers as well as ADNI 305_S_6744.
While a DateTime is included in the SOPInstanceUID for Baxter’s data, the variability suggests it is not encoding acquisition time (though neither does 0008,0033). For example, a 4 volume sequence with a TR = 1.5s (6s of acquisition) has 0008,0018 span 9.89s.
(0008,0018) UI [1.3.46.670589.11.17240.5.0.8932.2018052615291998092] # 52, 1 SOPInstanceUID
(0008,0023) DA [20180526] # 8, 1 ContentDate
(0008,0033) TM [152913.83] # 10, 1 ContentTime
The ADNI data does not embed the DateTime in the SOPInstanceUID:
(0008,0018) UI [2.16.124.113543.6006.99.8494861114665796968] # 44, 1 SOPInstanceUID
(0008,0023) DA [20190629] # 8, 1 ContentDate
(0008,0033) TM [094245.63] # 10, 1 ContentTime
For the instances I looked at, neither 0008,0018 nor 0008,0033 provide plausible slice times. Perhaps they reflect database or reconstructor times.
Perhaps this method is useful for some Philips data, but it does not generalize. I would also caution that simply subtracting times in YYYYMMDDHHmmSS format is prone to fault. For example, there are 60 seconds in a minute so the mSS will go 058,059,100,101… Similar issues for hours and running across midnight.
Also, once upon a time Philips stored data to disk in the same order they were acquired. This is no longer the case, and I speculate the order reflects when images are returned from a massively parallel pipeline. For Philips enhanced data you want to look at DimensionIndexValues to discriminate between temporal volumes (fMRI), directions (DWI), slices and echoes.
(0020,9057) UL 1 # 4, 1 InStackPositionNumber
(0020,9128) UL 1 # 4, 1 TemporalPositionIndex
(0020,9157) UL 1\1\1\1 # 16, 4 DimensionIndexValues
For individuals acquiring their own data, I encourage directly measuring slice time by intentionally moving your head during a volume to reveal slice order.
Thank you so much for checking ADNI data.
It is so complicated to get slice timing from ADNI Philips data.
Do you think it is better to get slice order from the ADNI data description dictionary, and then calculate the slice timing as you previously explained?
I have not actively used Philips equipment since 2005, so my knowledge is very out of date. Current Philips users are better placed than me to answer this. I would encourage scientists with Philips systems to lobby their Research Collaboration Managers to include details like phase encoding polarity and slice timing into future DICOM images.
With that caveat, and without the physical validation data I suggested previously, your best bet is to look for the details provided for the dataset. For example, the ADNI axial fcMRI sequence suggests equidistant Temporal slice spacing, prospect. motion corr. is off (if on, a short delay occurs between volumes), TR is 3.0s, 48 slices, 1 package, and default Slice scan order. The default slice order for a single package is described here.
MANAGED BY INCF