Repository navigation
What's Changed
- 8971 Skip Tests On Docs Only by @ericspod in #9109
- fix(utils): confine safe_eval name resolution to caller-supplied variables by @rubenG1009 in #9125
- Fix diffusion inferers to support diffusers-style schedulers by @ugbotueferhire in #8870
- fix(transforms): preserve device and dtype in squarepulse by @aymuos15 in #9005
- tests(inferers): remove dead TEST_CASE_ERROR_10 variable by @aymuos15 in #9031
- tests: remove byte-identical duplicate test cases by @aymuos15 in #9051
- Fix error messages missing the f-string prefix by @Yehsin0815 in #9139
- Expand ONNX export test coverage to 12 network architectures by @chhayankjain in #9073
- Perf(test): shrink synthetic volumes in image loading and dataset tests by @aymuos15 in #9034
- Perf(test): reuse random data across container types in quantile test by @aymuos15 in #8985
- perf(test): cache SwinUNETR SSL weights instead of re-downloading by @aymuos15 in #8987
- perf(RandCropBoxByPosNegLabeld): avoid N-fold mask allocation in randomize by @aymuos15 in #8993
- Fix argument names in docstrings by @caitlin-leonard in #9149
- Perf: single-pass remap_instance_id (unique + bincount + LUT gather) by @aymuos15 in #9009
- Fix RandGridDistortiond crash when transform is skipped by @HeyangQin in #8788
- fix(transforms): resolve TypeError when spatial_size is None (#9068) by @FinalSunFlower in #9080
- fix(data): skip empty foreground when computing DatasetSummary statistics by @aymuos15 in #9043
- fix(losses): BarlowTwinsLoss preserves float16/bfloat16 input dtype by @tritsystem in #9103
- Handle constant B-spline mutual information inputs by @yinkev in #9022
- Record system resource usage with MLFlowHandler by @uditmahato in #9052
- Accept torch tensors as DiNTS arch_code for weights_only=True checkpoints by @minsuking in #9053
- Avoid torch.compile graph breaks on DiNTS topology branches by @nilapate in #9145
- Fix RecursionError in set_rnd on cyclic object graphs (#8087) by @ousamabenyounes in #9056
- Fix Warp reference grid caching by @JESUSROYETH in #9082
- Add ROCm support for AMD Instinct GPUs by @nilapate in #9153
- fix(metrics): score a missed prediction in Hausdorff percentile, don't drop it by @asifuddin01 in #9096
- [pre-commit.ci] pre-commit suggestions by @pre-commit-ci[bot] in #9154
- fix(metrics): stop passing deprecated always_return_as_numpy internally (Fixes #9059) by @venki-drn in #9060
- Add average calibration losses by @theo-barfoot in #9092
- Make NormalizeIntensity invertible (#5647) by @azrabano23 in #8905
- Fix GridPatchDataset cache edge cases (#9100) by @QiuLsG in #9101
- Accept tensors and lists for DiNTS arch_code (fixes #9025) by @12yuuuu in #9143
- Add instance ID support to synthetic test images by @2260012471-eng in #9102
- Fix FID metric for scipy >= 1.18: sqrtm no longer accepts disp by @dhillrigo in #9076
- Fix ZarrAvgMerger under zarr 3.3: map chunks=True to auto-chunking by @dhillrigo in #9077
- Fix Spacingd updating non-metadata entries that share the key prefix by @garciadias in #9167
- fix(networks): AffineHead follows the module's dtype, not just device by @tritsystem in #9104
- Adding Deprecation Warning For hash_type by @ericspod in #9130
- feat(transforms): add MarchingCubes surface extraction by @aymuos15 in #9107
- [Fixes #9116] Include per-class rank in EmbeddingCollapseMetric aggregate score by @DucAnhValentinoNguyen in #9117
- Fix MeanIoU ignore_index to exclude voxels, not just a channel by @SID-6921 in #9134
- Adding per_component functionality to Surface Distance metric by @VijayVignesh1 in #8855
- Deprecate MMAR APIs and disable load_from_mmar by @Talhax55z in #9179
- docs: add missing upsample_mode parameter documentation to SegResNetDS and SegResNetDS2 by @Talhax55z in #9131
- Add anisotropic edge-case tests for SegResNetDS shape validation by @Talhax55z in #9128
- feat(metrics): add likelihood ratios and diagnostic odds ratio to ConfusionMatrixMetric by @qinxwew in #9105
- Fix spatial_ndim after EnsureChannelFirst adds a channel dim by @Spenhouet in #9185
- Remove docs for parameters that don't exist by @caitlin-leonard in #9169
- fix(ZarrAvgMerger): use separate tmpdir attrs to prevent value_store tmpdir GC on zarr v3 by @Zeesejo in #8816
- fix(metrics): return the minimum for Hausdorff percentile=0 by @aymuos15 in #9033
- Fix eager nvimgcodec plugin import by @vikasreddy11 in #9176
- fix(metrics): exclude ignored voxels from confusion matrix scores by @JoaoMarcos44 in #9195
- feat(networks): opt-in native-resolution deep supervision output for DynUNet by @aymuos15 in #9127
- Fix/intensity remap range by @devteamaegis in #9155
- Fix unspecified spatial size for rank-one resampling and add NIfTI regression tests by @moonbirdyang in #9191
- Fix per_component Dice silently ignoring ignore_index by @SID-6921 in #9171
- Check nnUNet plans network names against allowed modules by @garciadias in #9237
- Restore algo state values literally in algo_from_json by @garciadias in #9233
- Quote arguments in the NNI trial command by @garciadias in #9235
- Confine BundleAlgo.export_to_disk to the output directory by @garciadias in #9236
- Add ConvNeXt for 1D, 2D and 3D inputs by @VenkateswarluNagineni in #8995
- Add root-level AGENTS.md for AI coding assistants by @arxh80 in #9212
- Fix argument names in QuickNAT docstrings by @caitlin-leonard in #9148
- test(nifti): cover 2D writer without target affine by @px12811 in #9173
- Fix compiled lazy resampling with align_corners=True by @vikasreddy11 in #9248
- Fix setuptools csrc package warnings and include native sources in sdist by @vikasreddy11 in #9260
- Fix NormalizeIntensity inverse for non-contiguous inputs by @mottopanikeiku in #9183
- Keep argument boundaries when building Auto3DSeg commands by @garciadias in #9234
- Fix CSVIterableDataset yielding data twice when shuffle=True by @Mahdi-Hasan in #9249
- Treat BundleGen algos mapping values as literal data by @garciadias in #9253
- Do not evaluate imports when reading an algorithm score by @garciadias in #9254
- Factor LPIPS into MONAI perceptual loss by @oynal in #9287
New Contributors
- @Yehsin0815 made their first contribution in #9139
- @caitlin-leonard made their first contribution in #9149
- @HeyangQin made their first contribution in #8788
- @FinalSunFlower made their first contribution in #9080
- @tritsystem made their first contribution in #9103
- @uditmahato made their first contribution in #9052
- @nilapate made their first contribution in #9145
- @ousamabenyounes made their first contribution in #9056
- @JESUSROYETH made their first contribution in #9082
- @asifuddin01 made their first contribution in #9096
- @venki-drn made their first contribution in #9060
- @azrabano23 made their first contribution in #8905
- @QiuLsG made their first contribution in #9101
- @12yuuuu made their first contribution in #9143
- @2260012471-eng made their first contribution in #9102
- @dhillrigo made their first contribution in #9076
- @DucAnhValentinoNguyen made their first contribution in #9117
- @qinxwew made their first contribution in #9105
- @vikasreddy11 made their first contribution in #9176
- @JoaoMarcos44 made their first contribution in #9195
- @devteamaegis made their first contribution in #9155
- @moonbirdyang made their first contribution in #9191
- @arxh80 made their first contribution in #9212
- @px12811 made their first contribution in #9173
- @mottopanikeiku made their first contribution in #9183
- @Mahdi-Hasan made their first contribution in #9249
- @oynal made their first contribution in #9287
Full Changelog: 1.6.1...1.6.2