GE AcquisitionDuration about dcm2niix HOT 6 CLOSED

Please @neurolabusc review the proposal or the above commit. I will submit the PR once you gives okay.

from dcm2niix.

Hi @neurolabusc and @mr-jaemin,

If I may make a little suggestions, even though this is nice have this field in a human readable format, it is maybe not desired in every situation (and this is very GEHC specific).

As there is already some specific fields enable only when a certain compilation flag is enabled, maybe we can create a GEHC specific compile flag.

for example in

What do you think ? what that be reasonable ?

from dcm2niix.

My sense is to add these features to all version of dcm2niix, so different users do not get different details. I do think that we should not included redundant tags where they include the same informatin (AcquisitionDuration vs AcquisitionDurationGE), we should use private tags when they provide more precise information than public tags (0019,105A vs 0018,9073), and we should include sequence details in the JSON files if they may aid subsequent processing (EchoSpacingMicroSecondsGE , NotPhysicalNumberOfAcquiredPELinesGE and NotPhysicalTotalReadOutTimeGE). Happy to meet and build consensus if there is disagreement on these thoughts.

from dcm2niix.

@mr-jaemin I have two minor comments.

First, your premise is that the private tag 0019,105A is more accurate than the public tag 0018,9073. Therefore, I would ensure that the private tag always gets precedence. While tags are typically provided sequentially, this may not be the case for nested details. Therefore, I would change nii_dicom.cpp:

		case kAcquisitionDuration:
			if (!isSameFloatGE(d.acquisitionDuration, 0.0))
				break; //issue 808: give precedence to more precise measures
			//n.b. used differently by different vendors https://github.com/rordenlab/dcm2niix/issues/225
			d.acquisitionDuration = dcmFloatDouble(lLength, &buffer[lPos], d.isLittleEndian);
			break;

Second, I would not add the AcquisitionDurationGE field to nii_dicom_batch.cpp as it is completely redundant with AcquisitionDuration (though rounded so it stores precision).

If you implement my second comment, you will delete the sprintf() call from your pull request. I recognize that your usage of sprintf is safe, but please use snprintf() instead to remove distracting clang/LLVM compiler warnings. Specifically, clang will report warning: 'sprintf' is deprecated: This function is provided for compatibility reasons only. Due to security concerns inherent in the design of sprintf(3), it is highly recommended that you use snprintf(3) instead. While @yhuang43 and the FreeSurfer team have added a few sprintf() calls, they all require the USING_DCM2NIIXFSWRAPPER compilation flag and therefore do not impact the regular compilation.

from dcm2niix.

@neurolabusc Thanks for your comments.

• Good point to give precedence to more precise measures. Added some details

case kAcquisitionDuration:
   if (!isSameFloatGE(d.acquisitionDuration, 0.0))
      break; //issue 808: give precedence to more precise measures, e.g kAcquisitionDurationGE (0019,105a)

• Agreed that MM:SS format (AcquisitionDurationGE) is completely redundant with AcquisitionDuration in seconds.
  • I personally find this human readable format (MM:SS) quite useful to check the scan time. Will think/discuss more.
  • Regardless, thanks for the snprintf() suggestion (didn't know this compiler issue)
  • Any suggestion? C++ string like std:string vs C-style string like this example, in general?

from dcm2niix.

Agreed. Since both DICOM and BIDS AcquisitionDuration are in seconds, I would remove a redundant AcquisitionDurationGE (MM:SS) tag.

submitted PR #810

In summary,

• While GEHC doesn't report public tag (0018,9073) AcquisitionDuration yet,
• GEHC private tag (0019,105A) reports AcquisitionDuration in micro-seconds:
  (0019,105a) FL 3.0118515e+08 # 4, 1 AcquisitionDuration
• dcm2niix (v1.0.20240327+) extracts the private tag (0019,105A) which gets precedence over the (0018,9073)
• Report in the JSON files
  "AcquisitionDuration": 301.185,

from dcm2niix.