Flywheel Core 21.3

21.3.0

FEATURES

Clinical Trial Imaging Workflow Management

Flywheel 21.3.0 introduces comprehensive imaging workflow management capabilities designed to support clinical trial operations. These capabilities enable organizations to manage imaging data collection, quality control, and reader assessments across both research and regulated environments. This release includes the new Tasks Manager module integrated with the V3 Viewer, providing a flexible foundation for clinical imaging workflows with deployment options for non-regulated and 21 CFR Part 11 compliant environments.

Core Deployment Configuration

The following capabilities are available for all Flywheel Core deployments and provide essential tools for managing imaging workflows across research environments, including academic studies, biomarker development, algorithm testing, and clinical trial operations:

Tasks Manager

A new task management system enables teams to create, assign, and track work assignments across imaging studies. Key capabilities include:

• Task Creation and Assignment: Create individual or bulk tasks and assign them to specific users or staffing pools for flexible resource allocation. Tasks support custom protocols with structured data collection forms rendered in the viewer.
• Staffing Pools: Organize users into pools based on roles, expertise, or availability. Assign tasks to pools rather than individuals, allowing any qualified pool member to complete the work. This improves workload distribution and reduces bottlenecks in reader studies and imaging assessments.
• Task Filtering and Management: Filter and sort tasks by assignee, creator, due date, priority, status, and other attributes. Track task progress through statuses including To Do, In Progress, and Completed.
• Protocol-Based Workflows: Associate tasks with versioned protocols that define structured data collection forms. Forms are rendered directly in the V3 Viewer during task execution, ensuring consistent data capture across all tasks using the same protocol.
• Task Reporting: Generate CSV reports containing task metadata, form responses, and links to annotation files for external analysis. Export options include form and annotation reports (one row per response) and project management reports (one row per task). Project management reports include task lifecycle information such as task status changes, assignee, creator, due dates, and completion timestamps for operational oversight and analysis.

V3 Viewer Integration

The V3 Viewer has been enhanced to support task execution workflows:

• Task Execution Interface: Open and complete tasks directly within the viewer. View task details, fill out protocol-defined forms, and submit responses without leaving the viewing environment.
• Form Rendering: Protocol forms are dynamically rendered in the viewer based on JSON schemas. Forms support various question types and validation rules to ensure complete and accurate data collection.
• Annotations Management: Create and save image annotations (measurements, segmentations) as part of task completion. Annotations are linked to specific tasks and can be reviewed alongside form responses in the task details view.
• Draft Saving: Save tasks as drafts to preserve partial work, including form responses and annotations, and resume later. Draft tasks retain all in-progress data until the assignee is ready to complete submission.
• Viewer Configuration: Customize viewer behavior at the project level for all tasks within that project, including display settings, measurement tools, and annotation label colors. Project-level configuration ensures consistent viewing environments across all tasks in the project.

Validated Deployment Configuration

For organizations operating under 21 CFR Part 11 compliance requirements, Flywheel Validated Instance deployments include all Core capabilities above plus additional compliance features:

Electronic Signatures

Task completion workflows integrate with Flywheel's electronic signature system:

• Task Submission with Electronic Signature: When completing a task, users verify their identity with multi-factor authentication (MFA) and provide a customizable reason code documenting the intent of their submission. The system cryptographically signs all submitted data including form responses and annotations, creating an immutable record that links the authenticated user identity, submission timestamp, reason code, and all collected data. This non-repudiable signature ensures that the task submission cannot be altered after completion.
• Signature Verification: View signature details including signer identity, timestamp, and signature reason in completed task records. All signature events are recorded in the audit trail for regulatory traceability and inspection.

Audit Trail Integration

Task and form response activities are logged in the Flywheel audit trail:

• Form Response Logging: Each form submission is logged with details of who submitted responses, when the submission occurred, and the protocol version used. Form field values are tracked individually in the audit record.
• Annotation Logging: Annotation creation and updates during task execution are logged to the audit trail with timestamps and user information. Annotation files are linked to specific tasks for complete traceability.
• Project Locking: When projects are locked for validation or audit activities, all associated tasks, form responses, and annotations become read-only. This prevents any modifications to collected data during compliance reviews.

Guided Case Uploader

A streamlined interface for clinical trial sites to upload imaging data with step-by-step guidance that ensures data quality and consistency across all contributors.

Clinical trial imaging data often has non-human-readable file names, making it difficult for site staff to confirm they are uploading the correct files. The Guided Case Uploader addresses this challenge by automatically inspecting DICOM headers to display patient identifiers, study descriptions, and series information, allowing contributors to verify they have selected the correct data before submission.

Key capabilities include:

• Pre-Defined Upload Paths: Session Template validation guides contributors to upload data to consistent container paths (Subject, Session, Acquisition) across all sites participating in the trial, ensuring organizational consistency and enabling trial coordinators to locate data reliably.
• Supplementary Metadata Collection: Form-based data entry captures information beyond what is contained in the imaging files, such as site-specific notes, quality assessments, or protocol compliance indicators.
• DICOM Data Verification: Automatic inspection of DICOM headers displays PatientID, StudyInstanceUID, and SeriesDescription to help contributors confirm they have selected the correct files. The system detects inconsistencies across selected files and provides review tools to identify and resolve discrepancies before submission.
• Automatic File Organization: DICOM files are automatically grouped by series and organized into ZIP files, while non-DICOM files (TIFF, PNG) are handled appropriately within the same workflow.
• Bulk Import Pipeline Integration: All uploads are processed through Flywheel's Bulk Import pipeline, which enforces quality and security checks including malware scanning and conflict detection, and provides monitoring and traceability tools for trial coordinators and data managers.

Data Transfer

Expanded De-identification Support

Flywheel 21.3.0 expands de-identification capabilities across all data import workflows, providing organizations with comprehensive tools to protect patient privacy during data ingestion from multiple sources.

• De-ID for Imports from External Storage: Added de-identification support when importing data directly from External Storage Providers (AWS S3, Azure Blob Storage, Google Cloud Storage, or File System). Users can now apply de-identification profiles configured at the site, group, or project level during these imports.
• De-ID for Imports from Local Computer via CLI: Added de-identification support in the New CLI (fw-beta) for local computer imports. Data is now de-identified on the user's local computer before upload, matching the capability already available in the web browser import workflow. Previously, the New CLI did not perform any de-identification during import.
• De-ID Profile Testing Utility: Added a utility command in the New CLI (fw-beta) to test de-identification profiles locally before applying them during import. Users can now validate how their site, group, or project-level de-identification profiles will transform DICOM files, helping to ensure profiles are configured correctly before use.

User Attribution for Bulk Import

• When a user uploads and imports files from "My Computer" via Bulk Import, the files that eventually get placed into Core show the "Creator" as the user who uploaded the files.

Expanded Metadata & Data Type Support for Bulk Import

• Enhanced Bulk Import to allow setting project-level metadata fields from import mapping defaults, providing more flexibility in how data is imported and organized.
• Enhanced Bulk Import to support integer, float, and boolean data types in import mappings, in addition to strings. Values are automatically parsed as the appropriate type when mapping to custom metadata fields, preserving the original data types from imported datasets.

Error Handling Improvements

• Design improvements to more clearly indicate why the Bulk Import and Bulk Export jobs list is not shown, whether that is because the jobs list is still loading, an error occurred while fetching the jobs list, or there are in fact no jobs to display.
• The New CLI (fw-beta) debug mode is now available for all subcommands.
• Enhanced data export to automatically handle special characters in file and folder names that are not supported by local file systems. When exporting data, the system now substitutes problematic characters: asterisks (*) become star and all other filesystem-unsafe characters (such as colons, slashes, and vertical bars) are replaced with underscores (_), with consecutive underscores compacted to a single underscore. This prevents export failures on Windows, macOS, and Linux systems.
• Updated the export failure error dialog to use a modern UI component that aligns with the overall design of the application.

Usability Improvements

• Improved the Guided Case Uploader user experience with multiple enhancements including clearer upload progress indication, simplified cancel functionality, better control state management during uploads, and improved visual alignment and layout throughout the interface.
• Standardized file size display across the Flywheel interface for Bulk Import and Export to use SI units (base-10: KB = 1000 bytes) instead of binary units (base-2: KiB = 1024 bytes), improving consistency with operating systems and modern standards.

Viewers

V2 Viewer

• Performance optimization: added Cache-Control headers to V2 Viewer application bundle to reduce application load times.

V3 Viewer

• Added footer bar displaying instructions for the active tool.
• Improved error handling with descriptive error messages when the viewer configuration file is incorrect or when issues occur during viewer launch.
• Enhanced viewport synchronization: Zoom levels and pan can now be synchronized between viewports displaying the same dataset, even when the dataset is missing the ImageOrientationPatient DICOM tag.

Web App

Performance Improvements

• Faster page loads and navigation when accessing projects.
• Performance improvement to jobs log page loading by listing the most important data first.

FIXES

• Added a progress indicator for jobs log filtering.

Data Transfer

New CLI (fw-beta)

• Fixed installation issues on Windows 11, including correcting the PowerShell syntax for setting environment variables (adding required quotes around URLs) and replacing an unsupported console color that caused installation failures.
• Fixed an issue where creating filesystem storages on macOS could fail with confusing error messages due to symbolic link path resolution (for example, /tmp being a symbolic link to /private/tmp). The CLI now properly resolves symbolic link paths before validation.
• Fixed an issue where certain metadata fields (particularly those ending in "ts" or "time") were corrupted when migrating data between Flywheel instances using the --metadata flag.
• Fixed an issue where exporting data to a local filesystem path could fail with a TypeError.
• Improved error messages when import operations fail. The CLI now provides clear, descriptive error messages instead of generic AssertionError messages that did not explain what went wrong.
• Fixed an issue where the CLI would time out or fail when importing acquisitions containing large numbers of files (approximately 10,000 or more files per acquisition).

External Storages

• Fixed an issue where supplemental group IDs were not visible on the Create or Modify Storage Provider configuration panels, making it difficult to verify group access settings.
• Fixed an issue in the External Storage Provider configuration where values entered in the Supplemental Groups field could not be rearranged.

Bulk Import

• Fixed an issue where users with the Read-Write role could not view or resolve import conflicts for their own imports. Users with Read-Write access can now manage conflicts from imports they initiated, while users with Admin access can manage conflicts from any import in the project.
• Fixed an issue where imports could fail during malware scanning when datasets contained a large number of DICOM files. The maximum file limit for malware scanning was increased from 10,000 to 30,000 files to accommodate large fMRI and diffusion MRI datasets that can contain 25,000+ individual DICOM files.
• Fixed a display issue where de-identification profile validation errors would run together on a single line when multiple errors were present, making them difficult to read. Error messages now display properly formatted with line breaks between each validation issue.
• Fixed an issue where modifying a de-identification profile while an import job was in progress could result in inconsistent de-identification across files within the same import. The system now preserves the de-identification profile used at the start of each import job to ensure all files are processed with the same profile settings.

Bulk Export

• Improved error messaging when exports fail due to system interruptions. Previously, exports that failed due to system issues (such as insufficient disk space) displayed a misleading KeyboardInterrupt error message. The error message now clearly indicates "An unexpected error occurred" and provides guidance to retry or contact support.
• Fixed an issue where the folder browser would sometimes not appear when starting a Bulk Export to "My Computer" via the web app, preventing users from selecting the destination folder for their exported data.
• Fixed an issue where exporting archived files with the unzip option from older data snapshots would fail. The system now correctly retrieves the file contents list from the specified snapshot version rather than always using the latest version.
• Fixed an issue where bulk exports of large datasets would sometimes fail unexpectedly due to an internal server error.

Guided Case Uploader

• Fixed an issue where de-identification errors (such as attempting to increment empty date fields) caused valid DICOM files to be incorrectly identified as non-DICOM files. De-identification errors and DICOM header read errors are now properly distinguished.
• Fixed an issue where canceling a large series upload could inadvertently trigger multiple import operations. The uploader workflow was improved to prevent duplicate imports when canceling.
• Fixed an issue where de-identification errors on individual files could leave the upload in an inconsistent state.

Data views

• Resolve an issue with data views where adding the acquisition.label column to a view that already has the analysis.label column causes the view to return empty data.

Gears and Compute

• Resolved issue with FW compute that was causing long startup times for dynamic compute. This caused gear job startup times of 30 min in some cases.

Viewers

V2 Viewer

• Fixed indexing issues that could occur when rapidly creating multiple files in a single acquisition. This could happen when creating a Smart Copy of a project or during ingest of a large number of files.

21.3.1

FIXES

Data Management

• Fixed issue where multiple roles cannot be filtered in the "Role" column on the project dashboard.
• Fixed issue where space character cannot be typed into project search in project and group headers.

21.3.2

FIXES

Data Transfer

Legacy CLI (fw)

• Fixed an issue where the Legacy CLI (fw) ignored the --ssl-verify false flag, preventing connections to instances with self-signed certificates. The flag now works as documented, though users can also use the FW_CLI_SSL_VERIFY=false environment variable as an alternative.

New CLI (fw-beta)

• Fixed an issue where the New CLI (fw-beta) installation script on Windows would enter an infinite loop when the FW_SITE_URL environment variable was set. The installation now completes successfully in all cases.

Guided Case Uploader

• Improved Case Uploader usability by fixing button states during upload, preventing comment field values from carrying over between uploads, and correcting the upload progress time display to only show estimates after upload begins.
• Fixed an issue where Case Uploader file uploads could fail silently when incorrect authentication headers were sent to cloud storage. The uploader now shows clear error messages for failed uploads and only starts the import process when all files have been successfully uploaded.
• Fixed an issue where the Case Uploader could get stuck in an infinite loading loop when users lacked sufficient permissions to access groups or projects. The uploader now displays a clear error message instead of repeatedly attempting to reload.

Tasks Manager

• Fixed delete pool dialog to correctly display the count of tasks assigned to the specific pool being deleted, instead of showing the total count of all tasks.
• Fixed task list to maintain correct sorting order of the Created column when switching between manage mode and view mode.
• Fixed container label filtering in task list to correctly filter and display tasks based on selected container labels.
• Fixed task filtering by container label to correctly filter tasks based on the selected container label.
• Added search functionality to the project selector in the bulk task creation dialog, making it easier to find projects when users belong to many projects.
• Updated top bar title text to use Vision H4 typography style for consistent design across the application.
• Fixed protocol editor to correctly display "Draft v1" instead of "Draft v2" when editing the first draft of a newly created protocol configuration.
• Fixed spacing issue between the search bar and Create Pool button in the pool management interface.
• Improved the UI layout for Pool search.
• Fixed tasks to correctly launch on sites where multi-factor authentication is not enabled.

Viewers

V2 viewer

• Fixed an issue where image files that were changed from dicom type to archive were still showing in the Viewer.

21.3.3

FIXES

System Configuration

• Fixed configuration issue that made some menu options related to Tasks Manager visible on sites where it was not intended.

21.3.4

FIXES

Data Management

• Fixed an issue causing action menu items for subject and session lists to be intermittently disabled.

Reporting

• Fixed an issue causing usage data APIs to not collect and report data for recent data ranges.

Viewers

• Fixed a configuration issue that was hiding buttons to launch the default viewer at the session-level.

Guided case uploader

• Fixed an issue that caused data imported through the uploader to not be de-identified according to the project configuration.