PACS Desktop Viewer
When generating orthogonal image views from a series that contains an image not applicable to the volume, the viewer might attempt to use the inapplicable image if it processed the image before finding an applicable image, causing the viewer to crash.
The floating thumbnail panel’s header and footer labels failed to appear because it didn’t check the configuration settings when rendering the image.
If the RIS attempts to bookmark a study too quickly after launching a new viewer session, the viewer pump could get in a locked state, hanging the viewer.
After the viewer downloads the herpa data, it requests the image data. If the request was issued before the herpa data was completely processed, it could specify no data which put the stream server into a short sleep cycle, delaying the transfer of image data.
Images with referenced CAD SR objects listed in the thumbnail panel’s presentation group failed to load in an image frame properly because the CAD object doesn’t contain the necessary view information.
Loading a presentation state image into an image frame failed to apply the magnification and position settings because these characteristics are handled differently when loading the image into the image frame.
An unimplemented tool labeled Update Study was included in the default Session menu. It has been removed.
The processed orthogonal image doesn’t account for different horizontal and vertical pixel spacing and therefore cannot create a DICOM-compliant object that can be registered by the server. Uploading these objects has been disabled, which eliminates the crash. To create orthogonal view series that can be uploaded to the server and saved as presentation states and key images, generate the series using the MPR or 3D plugin extensions.
Updating the (frame) visibility state of an image while decompressing the image at the same time could lead to a deadlock because each process was its own thread.
The viewer was requesting the report data using a retired data format that the new json-formatted data didn’t support. As a result, the viewer was unable to display the report for review and approval.
Study loading performance would degrade as the number of images increased because larger image counts resulted in more high priority worker threads (to handle the larger number of IQ images). The number of high priority threads downgraded Windows UI refreshes making the viewer appear sluggish.
A timing issue when decompressing the final (zero-length) packet could result in a deadlock condition.
The change to json formatted data in the PBS file changed the way modality was encoded enabling the viewer to miss the modality value of the first study, which caused the viewer to select the wrong hanging protocol in mixed-modality studies.
When first loading a viewer session, an unnecessary delay might happen if the user does not update the images (e.g. scroll) in the image frame.
When using the viewer in the cloud environment and the study update notification appeared prompting the user to update the viewer session contents, the viewer didn’t use the cloud-specific function to restore the bookmarked session, resulting in an empty viewer session.
Some memory allocated when creating an overlay image was not released properly leading to a resource leak.
If image enhancement was applied to a series containing fewer than eight rows, the viewer could crash.
Loading a key image or presentation state into an image frame did not apply the saved position if the zoom factor and image position combined to create an image position outside the scope of the original image size.
If small thumbnails are used, the add study button did not appear on the unloaded study header, preventing users from loading the study. While the button icon might still be invisible, the function is invoked when the user clicks anywhere in the unloaded study’s study header.
If the viewer session starts with a hidden thumbnail panel, displaying it will place the study list on the second monitor due to a calculation error in the reset function.
The floating thumbnail panel failed to clear embedded calls to refresh the screen causing an infinite loop that slowed down the viewer scrolling function.
When adding a study to an active viewer session and creating the memory cache entry, the viewer failed to parse the study (PBS) data. As a result, the viewer failed to find the cache entry and have to recreate it on the fly.
An unintended change to the default rendering mode to one that used a sampling value smaller than the full quality. The original setting has been restored.
The viewer failed to set the priority when exporting images from a plugin allowing the memory manager to purge them before they were saved to the server.
A hanging protocol that included a fusion frame failed to record which functional series was loaded and always selected the first. Also, when the viewer loaded the hanging protocol, it searched for the first series that matched some criteria and stopped checking other series, which should have been selected because it might have matched the criteria better.
When loading a series functional series into a fusion frame, the viewer checks the series’s images to assure some part of the data can be used as the functional date. When the series contains no usable data, the viewer would crash rather than report the condition.
The JSON parser could fail to extract the accession number from the pbs data and exclude the study in the list sent to the WMI client. As a result, the WMI client believes the study was not loaded, causing subsequent Append or Restore commands to drop the study in the command.
If the network is slow to deliver an image or the initial quality compressed image size was less than the viewer needed, the image frame would display no data (image, overlay, etc.)
The Saige-Density CAD header didn’t display the error code when Saige processing failed.
When the viewer loaded a study and added images to a group (study), it could sort the data before it initialized the data values used to define the sort order. As a result of the incorrect sort order, the auto-splitting function would separate the images into additional series, causing unwanted series in the thumbnail panel.
In a multi-monitor environment, expanding the presentation group might not expand onto the second monitor.
When the thumbnail panel’s contents extended beyond the width of the thumbnail panel and the scroll position of the primary study’s header is off-screen, adding an object such as a key image or presentation state to the presentation group reset the primary study’s header position so it was again visible at the beginning of the thumbnail panel.
The coercion concatenation function, concat(), failed when the initial parameter in the instruction was a custom variable.
After issuing the record command, the viewer failed to check for engines using the conditional XML drop feature.
Dropping an XML file on a close event occurs only if an open or dictate event is configured.
When opening multiple studies into a viewer session, it was possible for the viewer to insert the wrong study identifiers in a dropped XML file because the viewer didn’t use the original study order when organizing the study list.
The XML file drop feature failed to create the file for built-in options if a condition was assigned to it.
If loading more cached studies into a viewer session than there are configured threads, the viewer might become deadlocked waiting for both images and presentation states processing tasks to complete.
There were cases in which present window/level setting labels were handled without regard to case, allowing the viewer to ignore some values. These labels are now all case sensitive.
There is an unnecessary lower limit set for supported image sizes and if an image did not contain enough data, the viewer crashed. The limit has been removed.
Plugin frames add their own window/level preset entries to the list but this didn’t happen under all conditions, resulting in missing options in the preset W/L list.
When loading a study into a viewer session, presentation state objects were not included in the prioritized download sequence. As a result, the thumbnail image would be unavailable until all other images were downloaded first.
If a rotated image is saved as a presentation state and loaded into an image frame, the auto-orientation setting reset the rotation setting, rendering the presentation state image without the rotation.
When the viewer is configured to drop multiple XML files for the same event, an uninitialized parameter could cause the viewer to mishandle the file type identifier, causing multiple files to have the same content.
If the user closed a viewer session before the study report view was initialized, the viewer would get stuck waiting for the report download to complete.
If the user closes the viewer session immediately after it launches, the viewer can check for the study state before it has set up the session fully, causing it to crash.
After uploading a presentation state fails and the user selects to save them locally, reopening the study loaded the saved objects into the viewer but didn’t (attempt to) upload them to the server.
When opening multiple studies, the tab position information was not properly reset, causing some tabs in the report panel to be missing or inaccessible.
Using the OB measurement annotation tools rather than the general measurement annotation tools to calibrate an image could result in a viewer crash because references to the calibration window were not cleared.
The format for storing spine labels changed causing the default spine line type value to change to spaces rather than vertebral bodies.
If the user reloaded an image into a frame after starting an annotation with presentation state generation set to auto-create, and then annotated the updated image, sending the presentation state to the server would fail and generate an invalid DICOM object.
If the cross-reference tool is applied to a CAD finding before the referenced image has been downloaded fully, the image could flicker in the frame.
When scrolling through a series in a tile view, only the first image tile was assured to be present. The remaining tile images, if not already downloaded, would appear empty (black) until the download completed.
When appending an uncooked study to a viewer session, the state and cooking progress information was not displayed, causing the viewer to appear hung until cooking completed.
Wrong tile view layout in image frame
When scrolling through a tiled image frame and the viewer came across an image that was not fully downloaded, no image was displayed, causing the tile to appear empty (black) until the image is fully downloaded.
When opening a viewer session containing a locked study with key images, and the user breaks the lock before all the images arrive, the viewer can crash when it updates the thumbnail list.
The viewer assumed a warning message received in a stream channel contained three values, but this was not always the case. As a result, it could mishandle the message.
When restoring a user profile, the viewer used the wrong data restoration pump causing error reporting and logging to fail. This could report the operation failed when it succeeded if the transfer took longer than expected, and failed if a custom port was defined.
When running a screen sharing application, such as Slack, the viewer’s drag-and-drop features failed to load a thumbnail in an image frame. The screen sharing app creates a transparent frame that the viewer did not recognize, preventing it from finding its target image frame.
The viewer failed to recognize some Windows messages used in Windows 11 to manipulate the window, such as maximizing the window by double-clicking the window title bar or dragging the title bar to the edge of the screen.
Deephealth MG CAD objects calculated the referenced frame number incorrectly, starting the index from zero rather than one. The viewer adopted Deephealth’s interpretation of referenced frame numbers, causing cross references of other MG CAD objects to be off by one frame. Note that the positioning of CAD findings and their markers was not affected. The interpretation has been resolved, as defined by DICOM, and existing MG CAD objects from other vendors are applied correctly. Existing Deephealth MG CAD objects containing the invalid index are handled as defined by HPV-218.
A failure to lock the stream task list could permit two streaming threads to update it at the same time, leading to a viewer crash when removing the stream task from the list after the herpa transfer completed.
After deleting the last shortcut macro from the shortcut manager panel, some buttons remained active but uninitialized, causing the viewer to crash if clicked.
A missing lock allowed the viewer to render a partially created key image thumbnail, causing the viewer to crash.
The viewer could crash when closing due to an unnecessary refresh of the thumbnail panel. Additional logging has been added to help identify the source of the thumbnail panel refresh.
Due to a long-standing, now obsolete display requirement, thin images (i.e., images with few columns) displayed in wide frames might not be justified to the frame edge when indicated, as when Smart Fit justifies the chest wall in mammography images to the frame edge. This restriction has been eliminated, allowing images of any width to be moved, manually or automatically, to any position in the image frame.
When uploading data from a DICOM CD containing a DICOMDIR file, encapsulated PDF objects were skipped and not uploaded because the upload application didn’t check for these object types.
NA
REVERSIBILITY NOTICE: If there are reports saved locally, due to a bookmarked study or crashed viewer session, these cannot be recovered if reverting to a prior version.
The viewer failed to check for an error after it submitted presentation states to the server and always reported the send completed successfully. The viewer now checks the status returned by the server. When there are errors, the message indicates which presentation states failed. The user can decide to continue without them or to delete the presentation state in the viewer and recreate it.
When using some language settings, the backup management panel failed to convert some characters to UTF-8 properly, resulting in invalid characters in place of accented characters in labels, dates and other text details.
When multiple studies reside on DICOM media created by eRAD PACS, the viewer might miscalculate the number of series or objects it’s been asked to open. As a result, only some of the series would be loaded.
When the user initiates a viewer session (Evo7 only) with the study currently being prefetched, the viewer might unlock the pump manager more than once, causing other threads to unlock prematurely and result in a hang or crash.
When the user interrupts image downloading, e.g., by closing a study while it is loading, the viewer fails to release some cache object references, leading to a resource leak. This has been resolved.
If the scheduler is interrupted when the CD viewer is loading images, the study being processed might not complete because the scheduler thread didn’t get restarted. As a result, some images would not be fully processed/decompressed.
If study load was interrupted when parsing the pbs file, the viewer might not free the study data container, leading to a memory resource leak.
If the call to draw an image failed to acquire a lock, it returned an error code the viewer didn’t recognize, resulting in inconsistent behavior.
For non-image data downloads, the temp pump was set to make a single request for data. If the data was greater than the requested size (32K), the remaining data was not downloaded.
An uninitialized lock age parameter could result in misleading data in a broken log message. To accommodate very old locks, the lock age variable was increased from 32-bits to 64-bits.
The viewer’s login dialog box was not started by the main thread. When the user logged in from the browser while this dialog was displayed, the viewer could hang or crash if they also attempted to enter data in the viewer’s login dialog box.
When the viewer resampled a thumbnail image using the image in memory (rather than from the herpa data), the viewer failed to set the thumbnail image size.
Links embedded on the settings panel failed when the applied language profile was not English because the links referenced the page by name, which changed after selecting another language.
The handshake between the cache manager and the viewer (specifically the study pump) could fail if the study pump is blocked but the pump manager removes a study from the list. The cache manager could remove the references after the study closes, but the viewer might attempt to use them when it opens the same study again.
The viewer failed to handle an exception, likely the result of a broken network connection, when logging in and out of a server, resulting in a viewer crash.
If the user received a popup notice while in the persistent recording state, the viewer could become stuck in that state after the popup clears and the user releases the record button.
When the viewer attempted to submit a report containing a status change, it could fail to include the impression data which would cause the server to hang, waiting for it. As a result, the viewer would get stuck in the closing state. This has been resolved.
When the viewer is running very slowly, for example, when a misconfigured antivirus solution is running, decompression, thumbnail creation and saving is delayed. As the viewer continues downloading the remaining compressed thumbnail data, the packets could get out-of-sync and some data can get released before it is used, causing the viewer to crash. This has been resolved.
The viewer failed to confirm a streaming task record still existed before attempting to access it. This has been resolved.
The scale at the bottom of the window/level distribution graph failed to recognize the greyscale photometric interpretation setting and always assumed zero was black, resulting in an inverted distribution graph when the image is defined as MONOCHROME1.
When using streaming protocol and the visible image frame in the original layout is set to use more tiles than there are images in the series rendered to that frame, the viewer would crash.
After the viewer stored the study UID in the stream server table, it incorrectly removed it when the session closed rather than when the study container was removed. If the viewer attempted to access the study referenced by the study container, it could crash.
The availability of a configurable user name and password to include in the XML file when dropped was applied to external reporting solutions with the specific label “XML Interface Speech Recognition” rather than any XML file drop solution.
When updating the thumbnail panel after receiving a request to add a study to the session, the viewer would allocate a structure to store the new studies, overriding the previous structure. However, the viewer did not properly remove the original structure, resulting in a read/write lock conflict and a deadlock condition.
If the reader thread and processing thread complete at the same time, the event signal can get crossed, causing one of the threads to miss the signal and hang indefinitely.
The locking mechanism could fail when the viewer attempted to allocate multiple read locks, causing a deadlock condition.
The application name is taken from the resource file but not until after the resource file was loaded. As a result, the application name was missing from the session log file.
If the herpa data is cached on the workstation and the open request is redirected to a server (for example, from a hub server to a child server when the open request was submitted to the worklist server), the viewer failed to initiate a function causing the study parsing to remain incomplete. And when the viewer session ended, the viewer would not close because it was waiting for the thread to close first.
If the thumbnail panel’s overlay information setting changed during or after closing the study, the pointer to the thumbnail header object became invalid and caused a crash when accessed.
After changing the grid mode setting to reuse data in populated image frames when eight or more image frames are on the screen with the selected frame being the last frame, and then changing the layout to one frame per screen, the viewer could crash because the viewer failed to update the selected image frame which no longer existed, even as a frame on a virtual monitor.
Some attempts have been made to address an unidentified crash in low-level Windows libraries when using the viewer’s patient folder.
If the server sent the viewer an invalid or empty stream server URL, an exception occurred but the viewer failed to catch it and crashed.
If the DICOM printer configuration contains an empty value for the true size flag setting, the viewer could crash when printing to the DICOM printer.
Spine labeling configuration data was mismanaged and the application of spine labels was mishandled, resulting in unavailable features or unintended behavior. This included the inconsistent storage of spine labels in the current XML-based storage format, improper application of spine label sets when modifying their configuration, and failure to update applied spine label values and connector lines after modifying their settings.
Continuing dictation recording after restoring a bookmarked session with a partial dictation popped up a progress panel that would not disappear after reloading completed.
Managing the streaming table failed to recognize a condition in which the study being removed from the table was also in the current sending list, resulting in a viewer crash.
After removing an auto-generated presentation state image, the viewer might still attempt to update it in the thumbnail panel, causing the viewer to crash.
When reloading a viewer session containing an image that was updated on the server, and for which a modality-specific preset W/L setting was applied, the viewer could crash because the viewer attempted to use the general preset values instead.
When populated virtual monitors exist, the viewer is configured to reuse populated frames, and the user sets the grid layout to some value twice, the second refresh displays no image in the first monitor’s frames.
When the LUT table provided in an image object contained zeroes in both the first and last position, calculating the minimum and maximum values resulted in an error causing the viewer to crash.
The viewer could crash when collecting an image’s decompression rate for the download percentage calculation because the thread was not locked.
A recent fix applied to restoring a bookmarked study resulted in the viewer blocking Windows application messages. One of these messages attempted to refresh a locked image, but the viewer didn’t receive the message.
If no data has started downloading before the plugin checks to see how much data has arrived, the plugin would crash.
After deleting an image or series that was exported from a plugin, the viewer failed to update the thumbnail panel. When the viewer refreshed the thumbnail panel, for any reason, the viewer would crash.
When the viewer is loading multiple studies and processing for the last two (or more) completes at the same time, the viewer can make multiple calls declaring the session is ready to use when only one is needed.
A failure to lock the streaming table when opening a study or appending a study to an existing session resulted in a crash.
Image files created by plugin modules remained in the viewer’s temp directory after uploading them to the server.
When creating a hanging protocol, the viewer failed to recognize the state of non-visible frames. As a result, series loading into non-visible frames when applying a hanging protocol displayed the incorrect viewed image icon on the thumbnail image.
A recent change eliminated a required thumbnail panel refresh. As a result, some data failed to appear in the thumbnail panel after appending a study to an active viewer session.
A change in the DICOM library resulted in a different return code when the viewer extracted the rescale slope value from the object. As a result, a default value wasn’t applied when no value was present, causing all pixel values to set to black.
The process for initializing large studies with many objects was not thread-safe, causing the viewer to crash when multiple threads attempted to initialize the same container.
An uninitialized modality LUT structure in the viewer could cause the viewer to crash if the image object defined a modality LUT.
Changes to the caching mechanism broke the DICOM processing mechanism in the standalone viewer, causing the standalone viewer to fail image loading. Note that the standalone viewer uses the thumbnail image until the full image has been fully processed and ready for display.
The fix for missing CAD markers when restoring a bookmarked viewer session disabled automatic application of CAD markers for stored data. As a result, the viewer failed to recognize the default CAD display setting saved in hanging protocols.
If the viewer received an open request, such as a second restore request, while still processing a bookmark restore request, the viewer could crash when the open request completed because it mismanaged the interrupted process.
When adding a key image to a prior study, as a report addendum, the viewer failed to submit it to the server because the mechanism for sending the key image had not been updated to make the call.
When the viewer-based scanning panel is opened from the Scan button on the worklist, some modules required the character set containers but they were not loaded, causing the viewer to crash.
If the print layout manager panel is invoked before the viewer has established a connection with the server, the viewer would crash.
The viewer was not checking the image data type so if it was the initial quality (IQ) image only and the remaining data did not follow (for some reason), the viewer could mislabel the image as full resolution.
The viewer called a refresh for every decompression event, even for images that were not visible. The large number of events might cause the viewer to miss a refresh, causing the image in the thumbnail panel to remain unchanged. Now, refresh events are called for visible images only.
The compression ratio overlay field misreported its value because the calculation was not using the full uncompressed image size.
After refactoring the image storing process, some data was not initialized when scanning an image, resulting in a crash.
When restoring an annotated, bookmarked image to an image frame, the viewer would incorrectly clear saved annotations like it does when temporary annotations exist.
Because the memory manager unnecessarily accessed the image data without locking it, it could use the image data while the viewer was closing the session. If the viewer released the image first, the viewer might crash.
When the streaming packet header is corrupted, data streaming stops and the channel needs to be reset.
When the viewer requested the performance measurement data from the server, it submitted the request on the data streaming connection. When the server submitted the response on the same connection, the data interfered with the streaming data, resulting in corrupted data. The viewer now requests performance data using the control stream connection.
The viewer failed to handle (empty) herpa data when the study consisted of non-image objects only, resulting in a crash. Now, the viewer displays the study details in the thumbnail header and indicates no image objects exist.
When prioritizing data for decompression, the viewer could download some non-visible images before fully downloading and decompressing visible images, causing a delay in rendering the visible images.
DEPENDENCY NOTICE: Dependencies exist. See details below.
DEPENDENCY NOTICE: This feature requires medsrv-9.0.2.
The herpa download uses streaming through the stream server rather than HTTP from the application server.
DEPENDENCY NOTICE: This feature requires medsrv-9.0.2.
The streaming technology has added support for gwav version 4, permitting better initial quality from smaller thumbnail images. The viewer still accepts gwav3 and gwav1, if offered by the server.
The UDI displayed in the viewer details panel has been set to 0086699400025590.
The baseline code base for version 9.0 viewer is viewer-8.0.85, desktopApp-8.0.31 and plugins-8.0.8 and includes the new (v9) caching mechanism (cache into memory, only), optimized image loading and only hyper+ streaming support.
All configurable settings and visible indications for and about retired features have been removed, including prefetch settings, cache management settings and HTTP-based streaming settings.
The series matching feature used by the hanging protocol manager and for series grouping has been optimized for performance, removing unnecessary calculations as well.
The viewer uses the stream server assigned by the server when streaming data during the session. If no stream server is provided, the viewer uses the server that sent the session (PBS) data.