RedBrick SDK Changelog

• Add project.upload.archive_tasks() for bulk archival of tasks.

• Move changelog to https://docs.redbrickai.com/changelogs/redbrick-sdk-changelog

• Add project.created_at and project.updated_at properties.

• Internal changes.

• Add org.delete_taxonomies() for bulk delete of taxonomies.

• Update org.taxonomies() to return complete list of taxonomies.
• Add org.delete_projects() for bulk delete of archived projects.

• API keys will be exported as API Key - {keyName}

• Improve concurrency of file downloads

• All exported label objects contain a labelId property, which can be directly used to create entity-level comments via project.upload.create_comment()

• Ability to add pins to global/entity comments:

"comment": {
  "text": <comment text>,
  "pin"?: {
    // coordinates are in image-space
    "pointX": <column>,
    "pointY": <row>,
    "pointZ": <slice index>,
    "frameIndex"?: <frame index for videos>,
    "volumeIndex"?: <series index>,
  }
}

• Internal API changes

• Enable webhook option

• Updated nibabel to force numpy v2
• Fixed update webhook API

• org.get_dataset() now returns RBDataset instead of Dict

• Export.list_tasks() to return tasks currently in the given stage if no search filter is specified

• DICOM Seg support in upload and export modules (--dicom-seg in CLI)

• Stage name validation in send_tasks_to_stage

• Update docs

• Create task comments - https://sdk.redbrickai.com/sdk.html#redbrick.common.upload.Upload.create_comment
• Add review_comment while reviewing tasks - https://sdk.redbrickai.com/sdk.html#redbrick.common.labeling.Labeling.put_tasks

• Update ModelStage.Config

• Silently bypass empty mask error on upload

• Added org.archive_project() / org.unarchive_project() to manage soft deletion.
• org.delete_project() now performs a full hard delete operation.

• Fix license check with tools like liccheck

• @marcus-wirtz-snkeos made their first contribution in https://github.com/redbrick-ai/redbrick-sdk/pull/214

• AltaDB Datasets

• Performance upgrade for semantic and binary mask exports

• OrgMember.is_active: bool
• Organization.Team.enable_members(member_ids: List[str]) -> None
• Organization.Team.disable_members(member_ids: List[str]) -> None Updated
• Organization.Team.list_members(active: bool = True) -> List[OrgMember]
• Workspace.update_datapoint_attributes(dp_id: str, attributes: Dict) -> None
• StorageMethod.AltaDB is now a RedBrick integrated storage method Removed
• Organization.Team.remove_members(member_ids: List[str]) -> None

• Use default series names in file paths

• Add storageId to task list and export

• Fix MHD segmentation direction

• Add labels export deserialization

• org.team.remove_members

• Support for multiple assignees

• Fix org.team.list_invites()

• team
• get_member(member_id: str) -> OrgMember
• list_members() -> List[OrgMember]
• remove_member(member_id: str) -> None
• list_invites() -> List[OrgInvite]
• invite_user(invitation: OrgInvite) -> OrgInvite
• revoke_invitation(invitation: OrgInvite) -> None
• storage
• get_storage(storage_id: str) -> StorageProvider
• list_storages() -> List[StorageProvider]
• create_storage(storage: StorageProvider) -> StorageProvider
• update_storage(storage_id: str, details: StorageProvider.Details) -> StorageProvider
• delete_storage(storage_id: str) -> bool
• verify_storage(storage_id: str, path: str) -> bool Workspace
• create_datapoints(storage_id: str, points: List[InputTask], concurrency: int = 50) -> List[Dict]
• delete_datapoints(dp_ids: List[str], concurrency: int = 50) -> bool
• update_datapoints_metadata(storage_id: str, points: List[Dict]) -> None
• add_datapoints_to_projects(project_ids: List[str], dp_ids: List[str], is_ground_truth: bool = False) -> None Project
• workforce
• get_member(member_id: str) -> ProjectMember
• list_members() -> List[ProjectMember]
• add_members(members: List[ProjectMember]) -> List[ProjectMember]
• update_members(members: List[ProjectMember]) -> List[ProjectMember]
• remove_members(member_ids: List[str]) -> None

• Unified control for client session

• Update consensus stage names

• Add status in task export

• Compress MHD segmentations

• project.upload.send_tasks_to_stage

• Dropped Python 3.8 support - https://endoflife.date/python
• Datapoint classification: https://sdk.redbrickai.com/formats/index.html#redbrick.types.task.OutputTask.datapointClassification
• MHD segmentations:
• Upload: https://sdk.redbrickai.com/sdk.html#redbrick.upload.Upload.create_datapoints & https://sdk.redbrickai.com/sdk.html#redbrick.upload.Upload.update_tasks_labels
• Label: https://sdk.redbrickai.com/sdk.html#redbrick.labeling.Labeling.put_tasks
• Export: https://sdk.redbrickai.com/sdk.html#redbrick.export.Export.export_tasks
• Also added prune_segmentations support in Label & Upload labels methods, which does a two-way pruning of uncommon instances in segmentMap and segmentations

• Upload attributes along with the datapoints
• Improve processing time of process_nifti_upload
• Fix bug in process_nifti_upload
• fix input types
• (v2.19.8) Fixes #197
• Make pydicom dep more flexible
• Download AltaDB series while exporting a taks
• export seriesFrameIndex only when it has a non null value
• Improve speed of binary mask exports

• Incorrect segmentation masks uploaded when multiple series and overlaps are present #197 proposed by @anaoum

• Fix workspace datapoint update query

• Fix bug in segmentations upload

• Improve processing time of process_nifti_upload

• 🎉 @anaoum made their first contribution in https://github.com/redbrick-ai/redbrick-sdk/pull/192

• Fixes segmentation uploads to azure containers.
• Fixes length measurement labels exports.
• Adds more model sub types to CT Segmentator.

• Add datapoint classification to export
• fix(redbrick/utils/rb_dicom_utils.py): Fix datapoint exports using SDK

• fix docker build dependencies

• Python 3.13

• Add python 3.13 support

• allow mixed formats task updates

• Patch an edge case in export of half-split data

• remove “test” from updated series name

• invalidate series items during update

• remove gzipping on upload
• Fields in label computed on label editor to be optional

• SDK changes for read only labels
• Update metadata along for tasks

• Fix consensus index issue

• add segmentMap[instanceId].overlappingGroups in export (docs)

• Set Consensus Settings during or after project creation

• https://sdk.redbrickai.com/formats/index.html#redbrick.types.task.Series.transforms
• https://sdk.redbrickai.com/formats/annotations.html#redbrick.types.task.CommonLabelProps.group

• redbrick.get_org_from_profile([profile])
• redbrick.get_project_from_profile([project_id], [profile])
• org.delete_project(project_id)
• org.delete_taxonomy([name], [tax_id])

• project.settings.task_duplication
• SDK Support for Local and External HeatMaps

• @AvanishCodes made their first contribution in https://github.com/redbrick-ai/redbrick-sdk/pull/178

• SDK Support for Local and External HeatMaps

• @AvanishCodes made their first contribution in https://github.com/redbrick-ai/redbrick-sdk/pull/178

• handle overlapping segmentations in rt-struct exports
• use asyncio.run instead of asyncio.get_event_loop

• REDBRICK_SDK_LOG_LEVEL=40 redbrick clone ...

import redbrick redbrick.config.log_level = 40 project = redbrick.get_project(…)

• Fix consensus segmentation export with empty base

• Boost CT Segmentator

• Minor patch

• Now available as a Docker image - https://hub.docker.com/r/redbrickai/redbrick-sdk
• Project creation improvements
• Project Webhooks
• Update task labels at any point in pipeline
• Workspaces support
• PNG binary mask upload

• Export only final labels

• Improve task export performance

• Trigger readthedocs build

• Docs improvements

• More docs improvements

• Update docs links

• Fix docs build

• update docstrings and readthedocs for formats by @shivam124081 🎉

• Add taxonomy types

• Fix label export index

• Global redbrick.config

• Add types for upload and export

• Export improvements

• Improved support for RT-Struct

• context.config.verify_ssl for upload/download files

• Add context.config

• Check global segmentations during upload

• TASK_CORRECTED event

• Advanced project creation
• Improved labeling metrics
• Import DICOM RT-Struct segmentations
• Use rt_struct=True in project.upload.create_datapoints() OR --rt-struct in redbrick upload
• The keys of segmentMap will be ROI names instead of instance ids -> {"Segment_1": "category"}

• Update Dependencies

• Chima/maj 2537 unit tests of sdkcli

• updated list_tasks assignee format

[{
    "taskId": str,
    "name": str,
    "createdAt": str,
    "updatedAt": str,
    "currentStageName": str,
    "createdBy"?: {"userId": str, "email": str},
    "priority"?: float([0, 1]),
    "metaData"?: dict,
    "series"?: [{"name"?: str, "metaData"?: dict}],
    "assignees"?: [{
        "user": str,
        "status": str,
        "assignedAt": datetime,
        "lastSavedAt"?: datetime,
        "completedAt"?: datetime,
        "timeSpentMs"?: float,
    }]
}]

• uploaded dicom files are stored in raw format

• Add .dicom support

• Feat/add tests
• Addresses #173

• @cAtaman made their first contribution in https://github.com/redbrick-ai/redbrick-sdk/pull/170

• Docs updates

• Bug fix (project clone)

• Logging improvements

• Add top-level series to consensus tasks

• --without-masks : Exports only tasks JSON without downloading any segmentation masks. Note: This is not recommended for tasks with overlapping labels.
• --semantic : Whether to export all segmentations as semantic_mask. This will create one segmentation file per class. If this is set to True and a task has multiple instances per class, then attributes belonging to each instance will not be exported.
• --binary-mask : Whether to export all segmentations as binary masks. This will create one segmentation file per instance.
• --single-mask: Whether to export all segmentations in a single file. Binary mask will be considered if both binary_mask and single_mask are set.

• without_masks: bool = False : Same as --without-masks
• without_json: bool = False : Doesn’t create the tasks JSON file.
• semantic_mask: bool = False : Same as --semantic
• binary_mask: Optional[bool] = None : True is same as --binary-mask and False is same as --single-mask

• Fix non-binary semantic class merging

• SDK: https://redbrick-sdk.readthedocs.io/en/stable/sdk.html#redbrick.export.Export.export_tasks

semantic_mask (bool = False) – Whether to export all segmentations as semantic_mask. This will create one instance per class. If this is set to True and a task has multiple instances per class, then attributes belonging to each instance will not be exported.

binary_mask (Optional[bool] = None) – Whether to export all segmentations as binary masks. This will create one segmentation file per instance. If this is set to None and a task has overlapping labels, then binary_mask option will be True for that particular task.

If both semantic_mask and binary_mask options are True, then one binary mask will be generated per class.

• CLI: redbrick export

• Fix typo

• Export cuboid labels

• Export groundtruth (c1431b9b519322c7aa0c1ff84429c987fb477c0a)
• Export consensus items (ea631b108c01370e2a78cb642d8cae954c7d19fc)
• Create project taxonomy (ae113dd77f2991e24af0e53179f31ceff9477185)

• Fix taxonomy format docs link

• Fix taxonomy format docs link

• Fix export series mapping

• Add sec. user ids to consensus export

• Fix ‘redbrick report’

• Internal server error on Export.get_active_time when task_id provided (#166)

• Unable to programatically determine the outcome of a create/update taxonomy operation (#165)

• Type distribution

• Implicit version checking (#163)

• SDK: redbrick.version()
• CLI: redbrick -v

• Fix segmentation export re-map issue

• Import NRRD images

• Support to export segmentations in DICOM RT-Struct format using --rt-struct in redbrick export / rt_struct: bool in Export.export_tasks
• Export.get_active_time

def get_active_time(
    self,
    *,
    stage_name: str,
    task_id: Optional[str] = None,
    concurrency: int = 100,
) -> Iterator[Dict]:
    """Get active time spent on tasks for labeling/reviewing.
    Parameters
    -----------
    stage_name: str
        Stage for which to return the time info.

    task_id: Optional[str] = None
        If set, will return info for the given task in the given stage.

    concurrency: int = 100
        Request batch size.

    Returns
    -----------
    Iterator[Dict]
        >>> [{
            "orgId": string,
            "projectId": string,
            "stageName": string,
            "taskId": string,
            "completedBy": string,
            "timeSpent": number,  # In milliseconds
            "completedAt": datetime,
            "cycle": number  # Task cycle
        }]
    """

• Export.list_tasks added param

completed_at: Optional[Tuple[Optional[float], Optional[float]]] = None
If present, will return tasks that were completed in the given time range.
The tuple contains the `from` and `to` timestamps respectively.

• Renamed RBOrganization.create_taxonomy_new -> RBOrganization.create_taxonomy
• Export.export_tasks now returns Iterator[Dict]
• Export.list_tasks now returns Iterator[Dict]
• Export.get_task_events now returns Iterator[Dict]
• Unified all user entities across exports to email of user.

• Export.search_tasks
• Export.redbrick_nifti
• Labeling.get_tasks
• Labeling.get_task_queue
• Labeling.assign_tasks (removed current_user param)

• Support DICOM RT-Struct segmentations export
• Export enhancements
• fix super truth export

• fix super truth export

• Support to export segmentations in DICOM RT-Struct format using --rt-struct in redbrick export / rt_struct: bool in Export.export_tasks
• Export.get_active_time

def get_active_time(
    self,
    *,
    stage_name: str,
    task_id: Optional[str] = None,
    concurrency: int = 100,
) -> Iterator[Dict]:
    """Get active time spent on tasks for labeling/reviewing.
    Parameters
    -----------
    stage_name: str
        Stage for which to return the time info.

    task_id: Optional[str] = None
        If set, will return info for the given task in the given stage.

    concurrency: int = 100
        Request batch size.

    Returns
    -----------
    Iterator[Dict]
        >>> [{
            "orgId": string,
            "projectId": string,
            "stageName": string,
            "taskId": string,
            "completedBy": string,
            "timeSpent": number,  # In milliseconds
            "completedAt": datetime,
            "cycle": number  # Task cycle
        }]
    """

• Export.list_tasks added param

completed_at: Optional[Tuple[Optional[float], Optional[float]]] = None
If present, will return tasks that were completed in the given time range.
The tuple contains the `from` and `to` timestamps respectively.

• Renamed RBOrganization.create_taxonomy_new -> RBOrganization.create_taxonomy
• Export.export_tasks now returns Iterator[Dict]
• Export.list_tasks now returns Iterator[Dict]
• Export.get_task_events now returns Iterator[Dict]
• Unified all user entities across exports to email of user.

• Export.search_tasks
• Export.redbrick_nifti
• Labeling.get_tasks
• Labeling.get_task_queue
• Labeling.assign_tasks (removed current_user param)

• Support DICOM RT-Struct segmentations export

• Fix image export for consensus tasks

• (v2.12.8) - Add superTruth in consensus export

• (v2.12.7) - Add extra info in Export.list_tasks

[{
    "taskId": str,
    "name": str,
    "createdAt": str,
    "currentStageName": str,
    "createdBy"?: {"userId": str, "email": str},
    "priority"?: float([0, 1]),
    "metaData"?: dict,
    "series"?: [{"name"?: str, "metaData"?: dict}],
    "assignees"?: [{"userId": str, "email": str}]
}]

• Upload.update_tasks_priority (v2.12.6)
• release v2.12.6

• Fix readthedocs (v2.12.5)

• Fix readthedocs (v2.12.4)

• Fix GH Actions (v2.12.3)

• Fix and update docs (v2.12.2)

• Update docs - Export.list_tasks

• Nested taxonomy and hints
• Labeling.update_tasks_priority + Series level meta data
• Misc improvements

• Labeling.update_tasks_priority + Series level meta data

• Nested taxonomy and hints

• Extensionless dicom files upload
• Update docs for enums
• Restrict project creation to use Taxonomy V2

• v2.11.1 - Fix project creation

• Workspace, tasks filter and major improvements
• alpha release
• list_tasks fix
• list_tasks fix
• release v2.11.0

• list_tasks fix

• list_tasks fix

• Workspace, tasks filter and major improvements
• alpha release

• improve handling file upload/download concurrency

• json handle dcm upload without extension

• Optimize non-segmentation task export

• v2.10.1 update packages

• Allow updating task items
• Support put labeling tasks with existing labels
• Add from_timestamp in task report export
• Export task report for all labeling cycles Fixed
• Memory management in nifti import/export processing
• Fast export for single task
• Export frameindex for DCM videos Deprecated
• Discontinue to_timestamp in task export

• v2.9.1 - Instance Classification

• Support Python 3.10 and 3.11

• Labeling interface changes

1. Labeling.assign_tasks

• Assign multiple tasks at once in the stage that they are currently in Changed:

1. Labeling.put_tasks

• All arguments except stage_name and tasks are keyword-only.
• finalize boolean indicates whether to submit or save the task as draft in the label stage
• review_result boolean indicates whether to accept or reject all tasks in the review stage

2. Labeling.get_task_queue

• Get tasks in users’ queue based on their email Removed

1. Labeling.assign_task

• In favour of Labeling.assign_tasks

• Update readthedocs theme, and docstrings.

• add ellipse and measurement stats to export

• CLI export images ‘—dicom-to-nifti’
• change task events report userId to email

• Remove rasterio dependency

• Generate task events report + Deprecate legacy export and import
• Add concurrency in upload
• better handle segmentation files with volume index without a label instance

• redbrick_nifti :: return-type labels
• update attribute format docs
• report command to fetch task events report
• deprecate legacy projects data import/export
• Add option for concurrency (Default: 50) in data upload
• Improve error feedback
• Compress all request and response data
• single series segmentation labels

• RB-1175 : Deprecate active learning
• Task Pre-Assignment
• fix nifti mask color for taxonomy v2

• project members list
• sdk support for new tax labels

• NIfTI png export, New Taxonomy

• Properly bump version

• Improvements to task import/export
• fix segmentation input type

• Fix export series index issue + Disable creating legacy project types

• Drop Python3.6 support
• redesign task format
• Consensus Tasks

• Drop Python3.6 support
• redesign task format

• simplify segmentation upload
• deleteTasks and new task object format

• simplify segmentation upload

• typo fix and support local paths when labelsPath is a list

• fix no labels export issue

• nifti - import/export overlapping labels

• fix readthedocs import issue and build

• support label from external storage on upload

• set label storage

• active learning + mask taxonomy colour + other improvements
• upgrade pillow==9.0.1 for python > 3.6
• cli export by stage

• fix data paginator iteration

• add ima to supported dicom extensions
• org.create_taxonomy + .nii, .nii.gz DICOM image upload

• task labeling time information per org

• add labeling methods to docs

• resurrect notebook event loop handler

• add methods to docs

• add task search and taxonomies methods

• sdk dicom upload with labels
• support custom storage id in upload + lint fixes
• Bump pillow from 8.3.2 to 9.0.0
• move task to start api
• pillow to 8.4.0 for python 3.6 support
• minor changes
• fix groundtruth tasks export
• export input labels

• skip task in export for all errors
• slicer: first draft
• remove slicer
• fix mask labels data type

• skip task in export for all errors
• slicer: first draft

• add cli export test
• segmentation export
• redbrick_nifti export
• workflow tests
• prep v1
• error handling and improvements
• CLI NIfTI export
• better file path and naming sync
• CLI upload
• polishing upload

• add exists_okay check to project creation
• automatically attach nest_asyncio when there is a running event loop
• make url an optional arg in main methods
• get task id for uploads
• CLI - I
• active learning info get only unassigned tasks
• cli export caching
• fix:: check dpId existence in cached dimensions
• minor improvement for cache validity
• Hotfix/segmentation export

• fix client error management

• use latestTaskData and labelsData for export

• @pritamrungta made their first contribution in https://github.com/redbrick-ai/redbrick-sdk/pull/50

• New method for checking processing state of training.

• bug fix: upload of segmentation masks with single class fixed.

• Connection request limiting and session re-use

• Allow get_learning_info call after start_processing

• Version bump for setup.py modifications.

• Rasterio has been removed as a dependency for windows machines. To upload/export png masks, windows users will have to manually install rasterio.
• Create datapoint with masks interface has been updated.

• datapoint.json file now uses task_id instead of dp_id.

• bug fix: correctly handle 0 label tasks in segmentation export.

• Variable naming changes.

• Bug fix: If RedBrick AI polygon segmentation holes/regions have only 2 vertices, ignore for PNG mask export.
• Feature: Fill holes on segmentation mask PNG export.

• Active learning upgrade
• Active learning v2

• minor bug fix for coco bbox area calculation

• Handle incorrect api key error