Merge tag '25.1.0'

25.1.0 (May 21, 2025)

New feature release in the 25.1.x series.

This release includes a small fix related to interpolations on the bounds
of the image field-of-view. This improves consistency with prior versions
of fMRIPrep for data with limited fields-of-view or small bounding boxes.

This release also adds support for providing fallback values for
``TotalReadoutTime`` metadata.
For Philips datasets, if ``EstimatedTotalReadoutTime`` or
``EstimatedEffectiveEchoSpacing`` are available, you can use these
by passing ``--fallback-total-readout-time estimated``.
For datasets with no readout time information, a numeric value can be passed,
for example, ``--fallback-total-readout-time 0.05``.
If the readout time information is known, it is preferable to encode directly
in the dataset, but there are cases where it is better to explicitly pass the
value to the software than inject uncertain metadata into the dataset.
In particular, the true readout time is not necessary for SyN-SDC, but a value
is nonetheless needed in order to estimate the inhomogeneity field.

This release is a long-term-support *candidate*. We will be performing
extensive tests and monitoring bug reports over the next couple of months
to determine whether issues can be addressed without making breaking changes.

Structural processing changes
-----------------------------

This release pins a version of sMRIPrep (0.18) that correctly handles precomputed
tissue probability maps.
The FAST segmentation also now skips an internal bias field correction step
that is redundant with ANTs N4 correction,
and may be inappropriate for non-human populations.

Fieldmap processing changes
---------------------------

This release pins a version of SDCFlows (2.13) that includes several improvements to
the SyN-SDC method.
In particular, the fieldmap prior that constrained the scope of displacements
was reintroduced and refined.

All merged pull requests
------------------------

* FIX: Use nearest mode for extrapolating data outside image boundaries (#3453)
* ENH: Replace c3d_affine_tool with a ConvertAffine interface (#3464)
* ENH: Add flag to fallback to Estimated* metadata or a passed value for TotalReadoutTime (#3423)

Author: effigies

CHANGES.rst

@@ -1,3 +1,53 @@
+25.1.0 (May 21, 2025)
+=====================
+New feature release in the 25.1.x series.
+
+This release includes a small fix related to interpolations on the bounds
+of the image field-of-view. This improves consistency with prior versions
+of fMRIPrep for data with limited fields-of-view or small bounding boxes.
+
+This release also adds support for providing fallback values for
+``TotalReadoutTime`` metadata.
+For Philips datasets, if ``EstimatedTotalReadoutTime`` or
+``EstimatedEffectiveEchoSpacing`` are available, you can use these
+by passing ``--fallback-total-readout-time estimated``.
+For datasets with no readout time information, a numeric value can be passed,
+for example, ``--fallback-total-readout-time 0.05``.
+If the readout time information is known, it is preferable to encode directly
+in the dataset, but there are cases where it is better to explicitly pass the
+value to the software than inject uncertain metadata into the dataset.
+In particular, the true readout time is not necessary for SyN-SDC, but a value
+is nonetheless needed in order to estimate the inhomogeneity field.
+
+This release is a long-term-support *candidate*. We will be performing
+extensive tests and monitoring bug reports over the next couple of months
+to determine whether issues can be addressed without making breaking changes.
+
+Structural processing changes
+-----------------------------
+
+This release pins a version of sMRIPrep (0.18) that correctly handles precomputed
+tissue probability maps.
+The FAST segmentation also now skips an internal bias field correction step
+that is redundant with ANTs N4 correction,
+and may be inappropriate for non-human populations.
+
+Fieldmap processing changes
+---------------------------
+
+This release pins a version of SDCFlows (2.13) that includes several improvements to
+the SyN-SDC method.
+In particular, the fieldmap prior that constrained the scope of displacements
+was reintroduced and refined.
+
+All merged pull requests
+------------------------
+
+* FIX: Use nearest mode for extrapolating data outside image boundaries (#3453)
+* ENH: Replace c3d_affine_tool with a ConvertAffine interface (#3464)
+* ENH: Add flag to fallback to Estimated* metadata or a passed value for TotalReadoutTime (#3423)
+
+
 25.0.0 (March 25, 2025)
 =======================
 New feature release in the 25.0.x series.

env.yml

@@ -13,33 +13,29 @@ dependencies:
   # git-annex for templateflow users with DataLad superdatasets
   - git-annex=*=alldep*
   # Base scientific python stack; required by FSL, so pinned here
-  - numpy=1.26
+  - numpy=2.2
   - scipy=1.15
-  - matplotlib=3.9
+  - matplotlib=3.10
   - pandas=2.2
   - h5py=3.13
   # Dependencies compiled against numpy, best to stick with conda
   - nitime=0.11
   - scikit-image=0.25
   - scikit-learn=1.6
   # Utilities
-  - graphviz=11.0
-  - pandoc=3.6
+  - graphviz=12.2
+  - pandoc=3.7
   # Workflow dependencies: ANTs
-  - ants=2.5
-  # 5.4.1 and 5.4.2 cause segfaults with ants
-  # Try to remove this ASAP
-  # https://github.com/conda-forge/ants-feedstock/issues/19
-  - libitk=5.4.0
+  - ants=2.6
   # Workflow dependencies: Connectome Workbench
   - connectome-workbench-cli=2.0
-  # Workflow dependencies: FSL (versions pinned in 6.0.7.13)
+  # Workflow dependencies: FSL (versions pinned in 6.0.7.17.20250415.fe1c582e)
   - fsl-bet2=2111.8
-  - fsl-flirt=2111.2
+  - fsl-flirt=2111.4
   - fsl-fast4=2111.3
   - fsl-fugue=2201.5
   - fsl-mcflirt=2111.0
-  - fsl-miscmaths=2203.2
+  - fsl-miscmaths=2412.4
   - fsl-topup=2203.5
   - pip
   - pip:

pyproject.toml

@@ -36,7 +36,7 @@ dependencies = [
     "requests >= 2.27",
     "sdcflows >= 2.13.0",
     "smriprep >= 0.18.0",
-    "tedana >= 23.0.2",
+    "tedana >= 25.0.0",
     "templateflow >= 24.2.2",
     "transforms3d >= 0.4",
     "toml >= 0.10",