1.4.0

Added

• Implemented Conjugate Gradient Solver to generate confidence maps. (#7876)
• Added norm parameter to ResNet (#7752, #7805)
• Introduced alpha parameter to DiceFocalLoss for improved flexibility (#7841)
• Integrated Tailored ControlNet Implementations (#7875)
• Integrated Tailored Auto-Encoder Model (#7861)
• Integrated Tailored Diffusion U-Net Model (7867)
• Added Maisi morphological functions (#7893)
• Added support for downloading bundles from NGC private registry (#7907, #7929, #8076)
• Integrated generative refactor into the core (#7886, #7962)
• Made ViT and UNETR models compatible with TorchScript (#7937)
• Implemented post-download checks for MONAI bundles and compatibility warnings (#7938)
• Added NGC prefix argument when downloading bundles (7974)
• Added flash attention support in the attention block for improved performance (#7977)
• Enhanced MLPBlock for compatibility with VISTA-3D (#7995)
• Added support for Neighbor-Aware Calibration Loss (NACL) for calibrated models in segmentation tasks (#7819)
• Added label_smoothing parameter to DiceCELoss for enhanced model calibration (#8000)
• Add include_fc and use_combined_linear argument in the SABlock (#7996)
• Added utilities, networks, and an inferer specific to VISTA-3D (#7999, #7987, #8047, #8059, #8021)
• Integrated a new network, CellSamWrapper, for cell-based applications (#7981)
• Introduced WriteFileMapping transform to map between input image paths and their corresponding output paths (#7769)
• Added TrtHandler to accelerate models using TensorRT (#7990, #8064)
• Added box and points conversion transforms for more flexible spatial manipulation (#8053)
• Enhanced RandSimulateLowResolutiond transform with deterministic support (#8057)
• Added a contiguous argument to the Fourier class to facilitate contiguous tensor outputs (#7969)
• Allowed ApplyTransformToPointsd to receive a sequence of reference keys for more versatile point manipulation (#8063)
• Made MetaTensor an optional print in DataStats and DataStatsd for more concise logging (#7814)

misc.

• Refactored Dataset to utilize Compose for handling transforms. (#7784)
• Combined map_classes_to_indices and generate_label_classes_crop_centers into a unified function (#7712)
• Introduced metadata schema directly into the codebase for improved structure and validation (#7409)
• Renamed optional_packages_version to required_packages_version for clearer package dependency management. (#7253)
• Replaced pkg_resources with the more modern packaging module for package handling (#7953)
• Refactored MAISI-related networks to align with the new generative components (#7989, #7993, #8005)
• Added a badge displaying monthly download statistics to enhance project visibility (#7891)

Fixed

transforms

• Ensured deterministic behavior in MixUp, CutMix, and CutOut transforms (#7813)
• Applied a minor correction to AsDiscrete transform (#7984)
• Fixed handling of integer weightmaps in RandomWeightedCrop (#8097)
• Resolved data type bug in ScaleIntensityRangePercentile (#8109)

data

• Fixed negative strides issue in the NrrdReader (#7809)
• Addressed wsireader issue with retrieving MPP (7921)
• Ensured location is returned as a tuple in wsireader (#8007)
• Corrected interpretation of space directions in nrrd reader (#8091)

metrics and losses

• Improved memory management for NACLLoss (#8020)
• Fixed reduction logic in GeneralizedDiceScore (#7970)

networks

• Resolved issue with loading pre-trained weights in ResNet (#7924)
• Fixed error where torch.device object had no attribute gpu_id during TensorRT export (#8019)
• Corrected function for loading older weights in DiffusionModelUNet (#8031)
• Switched to torch_tensorrt.Device instead of torch.device during TensorRT compilation (#8051)

engines and handlers

• Attempted to resolve the "experiment already exists" issue in MLFlowHandler (#7916)
• Refactored the model export process for conversion and saving (#7934)

misc.

• Adjusted requirements to exclude version 2.0 (#7859)
• Updated deprecated scipy.ndimage namespaces in optional imports (#7847, #7897)
• Resolved load_module() deprecation in Python 3.12 (#7881)
• Fixed Ruff type check issues (#7885)
• Cleaned disk space in the conda test pipeline (#7902)
• Replaced deprecated pkgutil.find_loader usage (#7906)
• Enhanced docstrings in various modules (#7913, #8055)
• Test cases fixing (#7905, #7794, #7808)
• Fix mypy issue introduced in 1.11.0 (#7941)
• Cleaned up warnings during test collection (#7914)
• Fix incompatible types in assignment issue (#7950)
• Fix outdated link in the docs (#7971)
• Addressed CI issues (#7983, #8013)
• Fix module can not import correctly issue (#8015)
• Fix AttributeError when using torch.min and max (#8041)
• Ensure synchronization by adding cuda.synchronize (#8058)
• Ignore warning from nptyping as workaround (#8062)
• Suppress deprecated warning when importing monai (#8067)
• Fix link in test bundle under MONAI-extra-test-data (#8092)

Changed

• Base Docker image upgraded to nvcr.io/nvidia/pytorch:24.08-py3 from nvcr.io/nvidia/pytorch:23.08-py3
• Change blossom-ci to ACL security format (#7843)
• Move PyType test to weekly test (#8025)
• Adjusted to meet Numpy 2.0 requirements (#7857)

Deprecated

• Dropped support for Python 3.8 (#7909)
• Remove deprecated arguments and class for v1.4 (#8079)

Removed

• Remove use of deprecated python 3.12 strtobool (#7900)
• Removed the pipeline for publishing to testpypi (#8086)
• Cleaning up some very old and now obsolete infrastructure (#8113, #8118, #8121)

New Contributors

• @alkamid made their first contribution in #7847
• @kephale made their first contribution in #7841
• @guopengf made their first contribution in #7875
• @dcfidalgo made their first contribution in #7984
• @K-Rilla made their first contribution in #7953
• @Han123su made their first contribution in #7934
• @mylapallilavanyaa made their first contribution in #7891
• @staydelight made their first contribution in #7769
• @ken-ni made their first contribution in #8055
• @borisfom made their first contribution in #7990
• @25benjaminli made their first contribution in #8057
• @bwittmann made their first contribution in #7969
• @slicepaste made their first contribution in #7814
• @che85 made their first contribution in #8091

Full Changelog: 1.3.2...1.4.0

1.3.2

Fixed

misc.

• Updated Numpy version constraint to < 2.0 (#7859)

1.3.1

Added

• Support for by_measure argument in RemoveSmallObjects (#7137)
• Support for pretrained flag in ResNet (#7095)
• Support for uploading and downloading bundles to and from the Hugging Face Hub (#6454)
• Added weight parameter in DiceLoss to apply weight to voxels of each class (#7158)
• Support for returning dice for each class in DiceMetric (#7163)
• Introduced ComponentStore for storage purposes (#7159)
• Added utilities used in MONAI Generative (#7134)
• Enabled Python 3.11 support for convert_to_torchscript and convert_to_onnx (#7182)
• Support for MLflow in AutoRunner (#7176)
• fname_regex option in PydicomReader (#7181)
• Allowed setting AutoRunner parameters from config (#7175)
• VoxelMorphUNet and VoxelMorph (#7178)
• Enabled cache option in GridPatchDataset (#7180)
• Introduced class_labels option in write_metrics_reports for improved readability (#7249)
• DiffusionLoss for image registration task (#7272)
• Supported specifying filename in Saveimage (#7318)
• Compile support in SupervisedTrainer and SupervisedEvaluator (#7375)
• mlflow_experiment_name support in Auto3DSeg (#7442)
• Arm support (#7500)
• BarlowTwinsLoss for representation learning (#7530)
• SURELoss and ConjugateGradient for diffusion models (#7308)
• Support for CutMix, CutOut, and MixUp augmentation techniques (#7198)
• meta_file and logging_file options to BundleWorkflow (#7549)
• properties_path option to BundleWorkflow for customized properties (#7542)
• Support for both soft and hard clipping in ClipIntensityPercentiles (#7535)
• Support for not saving artifacts in MLFlowHandler (#7604)
• Support for multi-channel images in PerceptualLoss (#7568)
• Added ResNet backbone for FlexibleUNet (#7571)
• Introduced dim_head option in SABlock to set dimensions for each head (#7664)
• Direct links to github source code to docs (#7738, #7779)

misc.

• Refactored list_data_collate and collate_meta_tensor to utilize the latest PyTorch API (#7165)
• Added str method in Metric base class (#7487)
• Made enhancements for testing files (#7662, #7670, #7663, #7671, #7672)
• Improved documentation for bundles (#7116)

Fixed

transforms

• Addressed issue where lazy mode was ignored in SpatialPadd (#7316)
• Tracked applied operations in ImageFilter (#7395)
• Warnings are now given only if missing class is not set to 0 in generate_label_classes_crop_centers (#7602)
• Input is now always converted to C-order in distance_transform_edt to ensure consistent behavior (#7675)

data

• Modified .npz file behavior to use keys in NumpyReader (#7148)
• Handled corrupted cached files in PersistentDataset (#7244)
• Corrected affine update in NrrdReader (#7415)

metrics and losses

• Addressed precision issue in get_confusion_matrix (#7187)
• Harmonized and clarified documentation and tests for dice losses variants (#7587)

networks

• Removed hard-coded spatial_dims in SwinTransformer (#7302)
• Fixed learnable position_embeddings in PatchEmbeddingBlock (#7564, #7605)
• Removed memory_pool_limit in TRT config (#7647)
• Propagated kernel_size to ConvBlocks within AttentionUnet (#7734)
• Addressed hard-coded activation layer in ResNet (#7749)

bundle

• Resolved bundle download issue (#7280)
• Updated bundle_root directory for NNIGen (#7586)
• Checked for num_fold and failed early if incorrect (#7634)
• Enhanced logging logic in ConfigWorkflow (#7745)

misc.

• Enabled chaining in Auto3DSeg CLI (#7168)
• Addressed useless error message in nnUNetV2Runner (#7217)
• Resolved typing and deprecation issues in Mypy (#7231)
• Quoted $PY_EXE variable to handle Python path that contains spaces in Bash (#7268)
• Improved documentation, code examples, and warning messages in various modules (#7234, #7213, #7271, #7326, #7569, #7584)
• Fixed typos in various modules (#7321, #7322, #7458, #7595, #7612)
• Enhanced docstrings in various modules (#7245, #7381, #7746)
• Handled error when data is on CPU in DataAnalyzer (#7310)
• Updated version requirements for third-party packages (#7343, #7344, #7384, #7448, #7659, #7704, #7744, #7742, #7780)
• Addressed incorrect slice compute in ImageStats (#7374)
• Avoided editing a loop's mutable iterable to address B308 (#7397)
• Fixed issue with CUDA_VISIBLE_DEVICES setting being ignored (#7408, #7581)
• Avoided changing Python version in CICD (#7424)
• Renamed partial to callable in instantiate mode (#7413)
• Imported AttributeError for Python 3.12 compatibility (#7482)
• Updated nnUNetV2Runner to support nnunetv2 2.2 (#7483)
• Used uint8 instead of int8 in LabelStats (#7489)
• Utilized subprocess for nnUNet training (#7576)
• Addressed deprecated warning in ruff (#7625)
• Fixed downloading failure on FIPS machine (#7698)
• Updated torch_tensorrt compile parameters to avoid warning (#7714)
• Restrict Auto3DSeg fold input based on datalist (#7778)

Changed

• Base Docker image upgraded to nvcr.io/nvidia/pytorch:24.03-py3 from nvcr.io/nvidia/pytorch:23.08-py3

Removed

• Removed unrecommended star-arg unpacking after a keyword argument, addressed B026 (#7262)
• Skipped old PyTorch version test for SwinUNETR (#7266)
• Dropped docker build workflow and migrated to Nvidia Blossom system (#7450)
• Dropped Python 3.8 test on quick-py3 workflow (#7719)

1.3.0

Added

• Intensity transforms ScaleIntensityFixedMean and RandScaleIntensityFixedMean (#6542)
• UltrasoundConfidenceMapTransform used for computing confidence map from an ultrasound image (#6709)
• channel_wise support in RandScaleIntensity and RandShiftIntensity (#6793, #7025)
• RandSimulateLowResolution and RandSimulateLowResolutiond (#6806)
• SignalFillEmptyd (#7011)
• Euclidean distance transform DistanceTransformEDT with GPU support (#6981)
• Port loss and metrics from monai-generative (#6729, #6836)
• Support invert_image and retain_stats in AdjustContrast and RandAdjustContrast (#6542)
• New network DAF3D and Quicknat (#6306)
• Support sincos position embedding (#6986)
• ZarrAvgMerger used for patch inference (#6633)
• Dataset tracking support to MLFlowHandler (#6616)
• Considering spacing and subvoxel borders in SurfaceDiceMetric (#6681)
• CUCIM support for surface-related metrics (#7008)
• loss_fn support in IgniteMetric and renamed it to IgniteMetricHandler (#6695)
• CallableEventWithFilter and Events options for trigger_event in GarbageCollector (#6663)
• Support random sorting option to GridPatch, RandGridPatch, GridPatchd and RandGridPatchd (#6701)
• Support multi-threaded batch sampling in PatchInferer (#6139)
• SoftclDiceLoss and SoftDiceclDiceLoss (#6763)
• HausdorffDTLoss and LogHausdorffDTLoss (#6994)
• Documentation for TensorFloat-32 (#6770)
• Docstring format guide (#6780)
• GDSDataset support for GDS (#6778)
• PyTorch backend support for MapLabelValue (#6872)
• filter_func in copy_model_state to filter the weights to be loaded and filter_swinunetr (#6917)
• stats_sender to MonaiAlgo for FL stats (#6984)
• freeze_layers to help freeze specific layers (#6970)

misc.

• Refactor multi-node running command used in Auto3DSeg into dedicated functions (#6623)
• Support str type annotation to device in ToTensorD (#6737)
• Improve logging message and file name extenstion in DataAnalyzer for Auto3DSeg (#6758)
• Set data_range as a property in SSIMLoss (#6788)
• Unify environment variable access (#7084)
• end_lr support in WarmupCosineSchedule (#6662)
• Add ClearML as optional dependency (#6827)
• yandex.disk support in download_url (#6667)
• Improve config expression error message (#6977)

Fixed

transforms

• Make convert_box_to_mask throw errors when box size larger than the image (#6637)
• Fix lazy mode in RandAffine (#6774)
• Raise ValueError when map_items is bool in Compose (#6882)
• Improve performance for NormalizeIntensity (#6887)
• Fix mismatched shape in Spacing (#6912)
• Avoid FutureWarning in CropForeground (#6934)
• Fix Lazy=True ignored when using Dataset call (#6975)
• Shape check for arbitrary types for DataStats (#7082)

data

• Fix wrong spacing checking logic in PydicomReader and broken link in ITKReader (#6660)
• Fix boolean indexing of batched MetaTensor (#6781)
• Raise warning when multiprocessing in DataLoader (#6830)
• Remove shuffle in DistributedWeightedRandomSampler (#6886)
• Fix missing SegmentDescription in PydicomReader (#6937)
• Fix reading dicom series error in ITKReader (#6943)
• Fix KeyError in PydicomReader (#6946)
• Update metatensor_to_itk_image to accept RAS MetaTensor and update default 'space' in NrrdReader to SpaceKeys.LPS (#7000)
• Collate common meta dictionary keys (#7054)

metrics and losses

• Fixed bug in GeneralizedDiceLoss when batch=True (#6775)
• Support for BCEWithLogitsLoss in DiceCELoss (#6924)
• Support for weight in Dice and related losses (#7098)

networks

• Use np.prod instead of np.product (#6639)
• Fix dimension issue in MBConvBlock (#6672)
• Fix hard-coded up_kernel_size in ViTAutoEnc (#6735)
• Remove hard-coded bias_downsample in resnet (#6848)
• Fix unused kernel_size in ResBlock (#6999)
• Allow for defining reference grid on non-integer coordinates (#7032)
• Padding option for autoencoder (#7068)
• Lower peak memory usage for SegResNetDS (#7066)

bundle

• Set train_dataset_data and dataset_data to unrequired in BundleProperty (#6607)
• Set None to properties that do not have REF_ID (#6607)
• Fix AttributeError for default value in get_parsed_content for ConfigParser (#6756)
• Update monai.bundle.scripts to support NGC hosting (#6828, #6997)
• Add MetaProperties (#6835)
• Add create_workflow and update load function (#6835)
• Add bundle root directory to Python search directories automatically (#6910)
• Generate properties for bundle docs automatically (#6918)
• Move download_large_files from model zoo to core (#6958)
• Bundle syntax # as alias of :: (#6955)
• Fix bundle download naming issue (#6969, #6963)
• Simplify the usage of ckpt_export (#6965)
• update_kwargs in monai.bundle.script for merging multiple configs (#7109)

engines and handlers

• Added int options for iteration_log and epoch_log in TensorBoardStatsHandler (#7027)
• Support to run validator at training start (#7108)

misc.

• Fix device fallback error in DataAnalyzer (#6658)
• Add int check for current_mode in convert_applied_interp_mode (#6719)
• Consistent type in convert_to_contiguous (#6849)
• Label argmax in DataAnalyzer when retry on CPU (#6852)
• Fix DataAnalyzer with histogram_only=True (#6874)
• Fix AttributeError in RankFilter in single GPU environment (#6895)
• Remove the default warning on TORCH_ALLOW_TF32_CUBLAS_OVERRIDE and add debug print info (#6909)
• Hide user information in print_config (#6913, #6922)
• Optionally pass coordinates to predictor during sliding window (#6795)
• Proper ensembling when trained with a sigmoid in AutoRunner (#6588)
• Fixed test_retinanet by increasing absolute differences (#6615)
• Add type check to avoid comparing a np.array with a string in _check_kwargs_are_present (#6624)
• Fix md5 hashing with FIPS mode (#6635)
• Capture failures from Auto3DSeg related subprocess calls (#6596)
• Code formatting tool for user-specified directory (#7106)
• Various docstring fixes

Changed

• Base Docker image upgraded to nvcr.io/nvidia/pytorch:23.08-py3 from nvcr.io/nvidia/pytorch:23.03-py3

Deprecated

• allow_smaller=True; allow_smaller=False will be the new default in CropForeground and generate_spatial_bounding_box (#6736)
• dropout_prob in VNet in favor of dropout_prob_down and dropout_prob_up (#6768)
• workflow in BundleWorkflow in favor of workflow_type(#6768)
• pos_embed in PatchEmbeddingBlock in favor of proj_type(#6986)
• net_name and net_kwargs in download in favor of model(#7016)
• img_size parameter in SwinUNETR (#7093)

Removed

• pad_val, stride, per_channel and upsampler in OcclusionSensitivity (#6642)
• compute_meaniou (#7019)
• AsChannelFirst, AddChanneland SplitChannel (#7019)
• create_multigpu_supervised_trainer and create_multigpu_supervised_evaluator (#7019)
• runner_id in run (#7019)
• data_src_cfg_filename in AlgoEnsembleBuilder (#7019)
• get_validation_stats in Evaluator and get_train_stats in Trainer (#7019)
• epoch_interval and iteration_interval in TensorBoardStatsHandler (#7019)
• some self-hosted test (#7041)

1.2.0

Added

• Various Auto3DSeg enhancements and integration tests including multi-node multi-GPU optimization, major usability improvements
• TensorRT and ONNX support for monai.bundle API and the relevant models
• nnU-Net V2 integration monai.apps.nnunet
• Binary and categorical metrics and event handlers using MetricsReloaded
• Python module and CLI entry point for bundle workflows in monai.bundle.workflows and monai.fl.client
• Modular patch inference API including PatchInferer, merger, and splitter
• Initial release of lazy resampling including transforms and MetaTensor implementations
• Bridge for ITK Image object and MetaTensor monai.data.itk_torch_bridge
• Sliding window inference memory efficiency optimization including SlidingWindowInfererAdapt
• Generic kernel filtering transforms ImageFiltered and RandImageFiltered
• Trainable bilateral filters and joint bilateral filters
• ClearML stats and image handlers for experiment tracking

misc.

• Utility functions to warn API default value changes (#5738)
• Support of dot notation to access content of ConfigParser (#5813)
• Softmax version to focal loss (#6544)
• FROC metric for N-dimensional (#6528)
• Extend SurfaceDiceMetric for 3D images (#6549)
• A track_meta option for Lambda and derived transforms (#6385)
• CLIP pre-trained text-to-vision embedding (#6282)
• Optional spacing to surface distances calculations (#6144)
• WSIReader read by power and mpp (#6244)
• Support GPU tensor for GridPatch and GridPatchDataset (#6246)
• SomeOf transform composer (#6143)
• GridPatch with both count and threshold filtering (#6055)

Fixed

transforms

• map_classes_to_indices efficiency issue (#6468)
• Adaptive resampling mode based on backends (#6429)
• Improve Compose encapsulation (#6224)
• User-provided FolderLayout in SaveImage and SaveImaged transforms (#6213)
• SpacingD output shape compute stability (#6126)
• No mutate ratio /user inputs croppad (#6127)
• A warn flag to RandCropByLabelClasses (#6121)
• nan to indicate no_channel, split dim singleton (#6090)
• Compatible padding mode (#6076)
• Allow for missing filename_or_obj key (#5980)
• Spacing pixdim in-place change (#5950)
• Add warning in RandHistogramShift (#5877)
• Exclude cuCIM wrappers from get_transform_backends (#5838)

data

• __format__ implementation of MetaTensor (#6523)
• channel_dim in TiffFileWSIReader and CuCIMWSIReader (#6514)
• Prepend "meta" to MetaTensor.__repr__ and MetaTensor.__str__ for easier identification (#6214)
• MetaTensor slicing issue (#5845)
• Default writer flags (#6147)
• WSIReader defaults and tensor conversion (#6058)
• Remove redundant array copy for WSITiffFileReader (#6089)
• Fix unused arg in SlidingPatchWSIDataset (#6047)
• reverse_indexing for PILReader (#6008)
• Use np.linalg for the small affine inverse (#5967)

metrics and losses

• Removing L2-norm in contrastive loss (L2-norm already present in CosSim) (#6550)
• Fixes the SSIM metric (#6250)
• Efficiency issues of Dice metrics (#6412)
• Generalized Dice issue (#5929)
• Unify output tensor devices for multiple metrics (#5924)

networks

• Make RetinaNet throw errors for NaN only when training (#6479)
• Replace deprecated arg in torchvision models (#6401)
• Improves NVFuser import check (#6399)
• Add device in HoVerNetNuclearTypePostProcessing and HoVerNetInstanceMapPostProcessing (#6333)
• Enhance hovernet load pretrained function (#6269)
• Access to the att_mat in self-attention modules (#6493)
• Optional swinunetr-v2 (#6203)
• Add transform to handle empty box as training data for retinanet_detector (#6170)
• GPU utilization of DiNTS network (#6050)
• A pixelshuffle upsample shape mismatch problem (#5982)
• GEGLU activation function for the MLP Block (#5856)
• Constructors for DenseNet derived classes (#5846)
• Flexible interpolation modes in regunet (#5807)

bundle

• Optimized the deepcopy logic in ConfigParser (#6464)
• Improve check and error message of bundle run (#6400)
• Warn or raise ValueError on duplicated key in json/yaml config (#6252)
• Default metadata and logging values for bundle run (#6072)
• pprint head and tail in bundle script (#5969)
• Config parsing issue for substring reference (#5932)
• Fix instantiate for object instantiation with attribute path (#5866)
• Fix _get_latest_bundle_version issue on Windows (#5787)

engines and handlers

• MLflow handler run bug (#6446)
• monai.engine training attribute check (#6132)
• Update StatsHandler logging message (#6051)
• Added callable options for iteration_log and epoch_log in TensorBoard and MLFlow (#5976)
• CheckpointSaver logging error (#6026)
• Callable options for iteration_log and epoch_log in StatsHandler (#5965)

misc.

• Avoid creating cufile.log when import monai (#6106)
• monai._extensions module compatibility with rocm (#6161)
• Issue of repeated UserWarning: "TypedStorage is deprecated" (#6105)
• Use logging config at module level (#5960)
• Add ITK to the list of optional dependencies (#5858)
• RankFilter to skip logging when the rank is not meeting criteria (#6243)
• Various documentation issues

Changed

• Overall more precise and consistent type annotations
• Optionally depend on PyTorch-Ignite v0.4.11 instead of v0.4.10
• Base Docker image upgraded to nvcr.io/nvidia/pytorch:23.03-py3 from nvcr.io/nvidia/pytorch:22.10-py3

Deprecated

• resample=True; resample=False will be the new default in SaveImage
• random_size=True; random_size=False will be the new default for the random cropping transforms
• image_only=False; image_only=True will be the new default in LoadImage
• AddChannel and AsChannelFirst in favor of EnsureChannelFirst

Removed

• Deprecated APIs since v0.9, including WSIReader from monai.apps, NiftiSaver and PNGSaver from monai.data
• Support for PyTorch 1.8
• Support for Python 3.7

1.1.0

Added

• Hover-Net based digital pathology workflows including new network, loss, postprocessing, metric, training, and inference modules
• Various enhancements for Auto3dSeg AutoRunner including template caching, selection, and a dry-run mode nni_dry_run
• Various enhancements for Auto3dSeg algo templates including new state-of-the-art configurations, optimized GPU memory utilization
• New bundle API and configurations to support experiment management including MLFlowHandler
• New bundle.script API to support model zoo query and download
• LossMetric metric to compute loss as cumulative metric measurement
• Transforms and base transform APIs including RandomizableTrait and MedianSmooth
• runtime_cache option for CacheDataset and the derived classes to allow for shared caching on the fly
• Flexible name formatter for SaveImage transform
• pending_operations MetaTensor property and basic APIs for lazy image resampling
• Contrastive sensitivity for SSIM metric
• Extensible backbones for FlexibleUNet
• Generalize SobelGradients to 3D and any spatial axes
• warmup_multiplier option for WarmupCosineSchedule
• F beta score metric based on confusion matrix metric
• Support of key overwriting in LambdaD
• Basic premerge tests for Python 3.11
• Unit and integration tests for CUDA 11.6, 11.7 and A100 GPU
• DataAnalyzer handles minor image-label shape inconsistencies

Fixed

• Review and enhance previously untyped APIs with additional type annotations and casts
• switch_endianness in LoadImage now supports tensor input
• Reduced memory footprint for various Auto3dSeg tests
• Issue of @ in monai.bundle.ReferenceResolver
• Compatibility issue with ITK-Python 5.3 (converting itkMatrixF44 for default collate)
• Inconsistent of sform and qform when using different backends for SaveImage
• MetaTensor.shape call now returns a torch.Size instead of tuple
• Issue of channel reduction in GeneralizedDiceLoss
• Issue of background handling before softmax in DiceFocalLoss
• Numerical issue of LocalNormalizedCrossCorrelationLoss
• Issue of incompatible view size in ConfusionMatrixMetric
• NetAdapter compatibility with Torchscript
• Issue of extract_levels in RegUNet
• Optional bias_downsample in ResNet
• dtype overflow for ShiftIntensity transform
• Randomized transforms such as RandCuCIM now inherit RandomizableTrait
• fg_indices.size compatibility issue in generate_pos_neg_label_crop_centers
• Issue when inverting ToTensor
• Issue of capital letters in filename suffixes check in LoadImage
• Minor tensor compatibility issues in apps.nuclick.transforms
• Issue of float16 in verify_net_in_out
• std variable type issue for RandRicianNoise
• DataAnalyzer accepts None as label key and checks empty labels
• iter_patch_position now has a smaller memory footprint
• CumulativeAverage has been refactored and enhanced to allow for simple tracking of metric running stats.
• Multi-threading issue for MLFlowHandler

Changed

• Printing a MetaTensor now generates a less verbose representation
• DistributedSampler raises a ValueError if there are too few devices
• OpenCV and VideoDataset modules are loaded lazily to avoid dependency issues
• device in monai.engines.Workflow supports string values
• Activations and AsDiscrete take kwargs as additional arguments
• DataAnalyzer is now more efficient and writes summary stats before detailed all case stats
• Base Docker image upgraded to nvcr.io/nvidia/pytorch:22.10-py3 from nvcr.io/nvidia/pytorch:22.09-py3
• Simplified Conda environment file environment-dev.yml
• Versioneer dependency upgraded to 0.23 from 0.19

Deprecated

• NibabelReader input argument dtype is deprecated, the reader will use the original dtype of the image

Removed

• Support for PyTorch 1.7

1.0.1

Fixes

• DiceCELoss for multichannel targets
• Auto3DSeg DataAnalyzer out-of-memory error and other minor issues
• An optional flag issue in the RetinaNet detector
• An issue with output offset for Spacing
• A LoadImage issue when track_meta is False
• 1D data output error in VarAutoEncoder
• An issue with resolution computing in ImageStats

Added

• Flexible min/max pixdim options for Spacing
• Upsample mode deconvgroup and optional kernel sizes
• Docstrings for gradient-based saliency maps
• Occlusion sensitivity to use sliding window inference
• Enhanced Gaussian window and device assignments for sliding window inference
• Multi-GPU support for MonaiAlgo
• ClientAlgoStats and MonaiAlgoStats for federated summary statistics
• MetaTensor support for OneOf
• Add a file check for bundle logging config
• Additional content and an authentication token option for bundle info API
• An anti-aliasing option for Resized
• SlidingWindowInferer adaptive device based on cpu_thresh
• SegResNetDS with deep supervision and non-isotropic kernel support
• Premerge tests for Python 3.10

Changed

• Base Docker image upgraded to nvcr.io/nvidia/pytorch:22.09-py3 from nvcr.io/nvidia/pytorch:22.08-py3
• Replace None type metadata content with "none" for collate_fn compatibility
• HoVerNet Mode and Branch to independent StrEnum
• Automatically infer device from the first item in random elastic deformation dict
• Add channel dim in ComputeHoVerMaps and ComputeHoVerMapsd
• Remove batch dim in SobelGradients and SobelGradientsd

Deprecated

• Deprecating compute_meandice, compute_meaniou in monai.metrics, in favor of
  compute_dice and compute_iou respectively

1.0.0

Added

• monai.auto3dseg base APIs and monai.apps.auto3dseg components for automated machine learning (AutoML) workflow
• monai.fl module with base APIs and MonaiAlgo for federated learning client workflow
• An initial backwards compatibility guide
• Initial release of accelerated MRI reconstruction components, including CoilSensitivityModel
• Support of MetaTensor and new metadata attributes for various digital pathology components
• Various monai.bundle enhancements for MONAI model-zoo usability, including config debug mode and get_all_bundles_list
• new monai.transforms components including SignalContinuousWavelet for 1D signal, ComputeHoVerMaps for digital pathology, and SobelGradients for spatial gradients
• VarianceMetric and LabelQualityScore metrics for active learning
• Dataset API for real-time stream and videos
• Several networks and building blocks including FlexibleUNet and HoVerNet
• MeanIoUHandler and LogfileHandler workflow event handlers
• WSIReader with the TiffFile backend
• Multi-threading in WSIReader with cuCIM backend
• get_stats API in monai.engines.Workflow
• prune_meta_pattern in monai.transforms.LoadImage
• max_interactions for deepedit interaction workflow
• Various profiling utilities in monai.utils.profiling

Changed

• Base Docker image upgraded to nvcr.io/nvidia/pytorch:22.08-py3 from nvcr.io/nvidia/pytorch:22.06-py3
• Optionally depend on PyTorch-Ignite v0.4.10 instead of v0.4.9
• The cache-based dataset now matches the transform information when read/write the cache
• monai.losses.ContrastiveLoss now infers batch_size during forward()
• Rearrange the spatial axes in RandSmoothDeform transforms following PyTorch's convention
• Unified several environment flags into monai.utils.misc.MONAIEnvVars
• Simplified __str__ implementation of MetaTensor instead of relying on the __repr__ implementation

Fixed

• Improved error messages when both monai and monai-weekly are pip-installed
• Inconsistent pseudo number sequences for different num_workers in DataLoader
• Issue of repeated sequences for monai.data.ShuffleBuffer
• Issue of not preserving the physical extent in monai.transforms.Spacing
• Issue of using inception_v3 as the backbone of monai.networks.nets.TorchVisionFCModel
• Index device issue for monai.transforms.Crop
• Efficiency issue when converting the array dtype and contiguous memory

Deprecated

• Addchannel and AsChannelFirst transforms in favor of EnsureChannelFirst
• monai.apps.pathology.data components in favor of the corresponding components from monai.data
• monai.apps.pathology.handlers in favor of the corresponding components from monai.handlers

Removed

• Status section in the pull request template in favor of the pull request draft mode
• monai.engines.BaseWorkflow
• ndim and dimensions arguments in favor of spatial_dims
• n_classes, num_classes arguments in AsDiscrete in favor of to_onehot
• logit_thresh, threshold_values arguments in AsDiscrete in favor of threshold
• torch.testing.assert_allclose in favor of tests.utils.assert_allclose

0.9.1

Added

• Support of monai.data.MetaTensor as core data structure across the modules
• Support of inverse in array-based transforms
• monai.apps.TciaDataset APIs for The Cancer Imaging Archive (TCIA) datasets, including a pydicom-backend reader
• Initial release of components for MRI reconstruction in monai.apps.reconstruction, including various FFT utilities
• New metrics and losses, including mean IoU and structural similarity index
• monai.utils.StrEnum class to simplify Enum-based type annotations

Changed

• Base Docker image upgraded to nvcr.io/nvidia/pytorch:22.06-py3 from nvcr.io/nvidia/pytorch:22.04-py3
• Optionally depend on PyTorch-Ignite v0.4.9 instead of v0.4.8

Fixed

• Fixed issue of not skipping post activations in Convolution when input arguments are None
• Fixed issue of ignoring dropout arguments in DynUNet
• Fixed issue of hard-coded non-linear function in ViT classification head
• Fixed issue of in-memory config overriding with monai.bundle.ConfigParser.update
• 2D SwinUNETR incompatible shapes
• Fixed issue with monai.bundle.verify_metadata not raising exceptions
• Fixed issue with monai.transforms.GridPatch returns inconsistent type location when padding
• Wrong generalized Dice score metric when denominator is 0 but prediction is non-empty
• Docker image build error due to NGC CLI upgrade
• Optional default value when parsing id unavailable in a ConfigParser instance
• Immutable data input for the patch-based WSI datasets

Deprecated

• *_transforms and *_meta_dict fields in dictionary-based transforms in favor of MetaTensor
• meta_keys, meta_key_postfix, src_affine arguments in various transforms, in favor of MetaTensor
• AsChannelFirst and AddChannel, in favor of EnsureChannelFirst transform

0.9.0

Added

• monai.bundle primary module with a ConfigParser and command-line interfaces for config-based workflows
• Initial release of MONAI bundle specification
• Initial release of volumetric image detection modules including bounding boxes handling, RetinaNet-based architectures
• API preview monai.data.MetaTensor
• Unified monai.data.image_writer to support flexible IO backends including an ITK writer
• Various new network blocks and architectures including SwinUNETR
• DeepEdit interactive training/validation workflow
• NuClick interactive segmentation transforms
• Patch-based readers and datasets for whole-slide imaging
• New losses and metrics including SurfaceDiceMetric, GeneralizedDiceFocalLoss
• New pre-processing transforms including RandIntensityRemap, SpatialResample
• Multi-output and slice-based inference for SlidingWindowInferer
• NrrdReader for NRRD file support
• Torchscript utilities to save models with meta information
• Gradient-based visualization module SmoothGrad
• Automatic regular source code scanning for common vulnerabilities and coding errors

Changed

• Simplified TestTimeAugmentation using de-collate and invertible transforms APIs
• Refactoring monai.apps.pathology modules into monai.handlers and monai.transforms
• Flexible activation and normalization layers for TopologySearch and DiNTS
• Anisotropic first layers for 3D resnet
• Flexible ordering of activation, normalization in UNet
• Enhanced performance of connected-components analysis using Cupy
• INSTANCE_NVFUSER for enhanced performance in 3D instance norm
• Support of string representation of dtype in convert_data_type
• Added new options iteration_log, iteration_log to the logging handlers
• Base Docker image upgraded to nvcr.io/nvidia/pytorch:22.04-py3 from nvcr.io/nvidia/pytorch:21.10-py3
• collate_fn generates more data-related debugging info with dev_collate

Fixed

• Unified the spellings of "meta data", "metadata", "meta-data" to "metadata"
• Various inaccurate error messages when input data are in invalid shapes
• Issue of computing symmetric distances in compute_average_surface_distance
• Unnecessary layer self.conv3 in UnetResBlock
• Issue of torchscript compatibility for ViT and self-attention blocks
• Issue of hidden layers in UNETR
• allow_smaller in spatial cropping transforms
• Antialiasing in Resize
• Issue of bending energy loss value at different resolutions
• kwargs_read_csv in CSVDataset
• In-place modification in Metric reduction
• wrap_array for ensure_tuple
• Contribution guide for introducing new third-party dependencies

Removed

• Deprecated nifti_writer, png_writer in favor of monai.data.image_writer
• Support for PyTorch 1.6