**Introduction** The [Brain Imaging Data Structure (BIDS)][1] recommends providing the tag [TotalReadoutTime][2] for echo planar imaging (EPI) images. This value is used by [FSL's TOPUP][3] tool. Spatial distortion is an inherent property of EPI images. However, if you acquire two spin-echo EPI images that are identical except for phase-encoding polarity, the two images should have distortion of the same magnitude but with opposite direction. TOPUP can leverage this by estimating the midpoint between these two images. In most cases, this is a completely optional field, as the [TOPUP][5] manual notes: > If your readout time is identical for all acquisitions you don't > neccessarily have to specify a valid value in this column (you can > e.g. just set it to 1), but if you do specify correct values the > estimated field will be correctly scaled in Hz, which may be a useful > sanity check. However, if you do wish to provide this value, you should be aware that TOPUP uses an definition of readout time, and the BIDS format expects this value to match the FSL definition. This is described in the [BIDS][6] format: > This is actually the "effective" total readout time , defined as the > readout duration, specified in seconds, that would have generated data > with the given level of distortion. It is NOT the actual, physical > duration of the readout train. If EffectiveEchoSpacing has been > properly computed, it is just EffectiveEchoSpacing * (ReconMatrixPE - > 1).3 . This parameter is REQUIRED if corresponding "field/distortion" > maps acquired with opposing phase encoding directions are present (see > 8.9.4). Therefore, sequence settings like Echo Spacing, AcquisitionMatrix, Partial Fourier, Parallel Reduction Factor In Plane (e.g. SENSE), Phase Oversampling and Interpolation need to be considered. Several formula have been [proposed][7] (as well as [here][8]) to determine TotalReadoutTime from Philips data. However, several of these formula make assumptions regarding field strength or sequence settings. Here we provide sample datasets that vary sequence settings to validate different formula. At present, we suggest that a robust formula is: ActualEchoSpacing = WaterFatShift / (ImagingFrequency * 3.4 * (EPI_Factor + 1)) TotalReadoutTIme = ActualEchoSpacing * EPI_Factor EffectiveEchoSpacing = TotalReadoutTime / (ReconMatrixPE - 1) WaterFatShift = 2001,1022 ImagingFrequency = 0018,0084 EPI_Factor = 0018,0091 or 2001,1013 ReconMatrixPE = 0028,0010 or 0028,0011 depending on 0018,1312 In order to validate this formula, one can request TOPUP to estimate the field in Hz using the [-fout][9] parameter. For example, turning on partial Fourier should not influence the field relative to a scan without this option. On the other hand, SENSE samples all of K-space more rapidly so should lead to less spatial distortion. The included Python script "eddy_hz.py" will apply Eddy to pairs of images acquired at [National Institute of Mental Health and Neurosciences (NIMHANS)][10]. To run this script you must first convert the DICOM images to the NIfTI format TOPUP expects, and the NIfTI image filename must be set to the protocol name (e.g. "dcm2niix -z y -f %p ~/DICOMS"). Next, one runs the script 'eddy_hz.py' to run TOPUP for each image. Finally, the script 'figures.py' can be run to generate calibrated plots (you will need to edit this script to point to a copy of [MRIcroGL][11]). Ideally, one would hope that the field map (in Hz) would be identical across settings. While these look close, it is unclear whether variability in the images below reflects the precision of the formula or the scanner shimming across these settings. The fact that the images differ in both shape as well as intensity suggests the latter is at least a factor. Baseline Image (Full Fourier) ![enter image description here][12] 0.619 Partial Fourier ![enter image description here][4] Partial FOV ![enter image description here][13] x2 Interpolation ![enter image description here][14] SENSE x2 ![enter image description here][15] SENSE x4 ![enter image description here][16] **Notes** * FSL's definition of TotalReadoutTime is related to their definition of [EffectiveEchoSpacing][17], which the FSL FEAT tooltip describes as "Effective EPI echo spacing ... is the time between echoes in successive k-space lines. If you are using an accelerated sequence (parallel imaging) then the number you need here is the echo spacing for the acquired lines divided by the acceleration factor" * Safal Khanal describes [numerous][18] recipes for calculating TotalReadoutTime. * These notes from the [University of Oregon][19] are clear, but do not describe sitations of acceleration, interpolation and phase oversampling. * Michael Harms developed a pragmatic approach to infer TOPUP's ambiguous use of TotalReadoutTime for Siemens scanners. We adopt this approach here. He acquired [images][20] in a single series with different settings for partial Fourier, SENSE, Phase Oversampling, etc., and corrected each with TOPUP. Crucially, he had TOPUP generate its estimated Fieldmap, which should be static across these. This allowed infer the [definitions][21] as 1. EffectiveEchoSpacing` = 1/[BWPPPE * ReconMatrixPE] 2. TotalReadoutTime = EffectiveEchoSpacing * (ReconMatrixPE - 1) [1]: https://bids.neuroimaging.io [2]: https://bids-specification.readthedocs.io/en/stable/04-modality-specific-files/01-magnetic-resonance-imaging-data.html#task-including-resting-state-imaging-data [3]: https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/topup/TopupUsersGuide [4]: https://files.osf.io/v1/resources/xvguw/providers/osfstorage/5e78eebd4a60a506ccbb1dfe?mode=render [5]: https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/topup [6]: https://bids-specification.readthedocs.io/en/stable/04-modality-specific-files/01-magnetic-resonance-imaging-data.html#task-including-resting-state-imaging-data [7]: https://neurostars.org/t/consolidating-epi-echo-spacing-and-readout-time-for-philips-scanner/4406 [8]: https://github.com/rordenlab/dcm2niix/issues/377 [9]: https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/topup/TopupUsersGuide#A--fout [10]: https://mbial.weebly.com [11]: https://github.com/rordenlab/MRIcroGL12/releases [12]: https://files.osf.io/v1/resources/xvguw/providers/osfstorage/5e78ed8c0cd06c06d20043d4?mode=render [13]: https://files.osf.io/v1/resources/xvguw/providers/osfstorage/5e78edbd4a60a506ccbb1b7b?mode=render [14]: https://files.osf.io/v1/resources/xvguw/providers/osfstorage/5e78ede04a60a506b7bb4794?mode=render [15]: https://files.osf.io/v1/resources/xvguw/providers/osfstorage/5e78ee4e0cd06c06cc00651f?mode=render [16]: https://files.osf.io/v1/resources/xvguw/providers/osfstorage/5e78ee810cd06c06ca005049?mode=render [17]: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1404&L=FSL&D=0&P=603392 [18]: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;c75b0537.1710 [19]: https://lcni.uoregon.edu/kb-articles/kb-0003 [20]: https://github.com/neurolabusc/dcm_qa/tree/master/In/TotalReadoutTime [21]: https://github.com/rordenlab/dcm2niix/issues/130#issuecomment-331608725