Medical Image Labeler

Interactively explore, label, and publish animations of 2-D or 3-D medical image data

Since R2022b

Description

The Medical Image Labeler app enables you to label ground truth data in medical images. Using the app, you can:

• Import multiple 2-D images or 3-D image volumes.

• Explore images as slice planes or volumes with anatomical orientation markers and scale bars. Navigate volume slices using crosshairs.

• Publish 2-D and 3-D PNG or PDF image snapshots and GIF animations.

• Create multiple pixel label definitions to label regions of interest. Label pixels using automatic algorithms such as flood fill, semi-automatic techniques such as interpolation, and manual techniques such as painting by superpixels.

• Connect to the Medical Open Network for AI (MONAI) Label platform to apply fully automated and interactive deep learning models for segmenting radiology images. This functionality requires the Medical Imaging Toolbox™ Interface for MONAI Label Library support package.

• Write, import, and use your own custom automation algorithm to automatically label ground truth data.

• Export the labeled ground truth data as a groundTruthMedical object. You can use this object to share labels with colleagues or for training semantic segmentation deep learning networks.

The Medical Image Labeler app supports 2-D images and image sequences stored in the DICOM format, as well as the formats in the imformats registry. An image sequence is a series of images related by time, such as ultrasound data. You can also import 2-D images or image series from medicalImage objects in the workspace. The app supports 3-D image volume data stored in the DICOM (single or multifile volume), NIfTI, NRRD, and TIFF file formats. You can also import 3-D image volumes from medicalVolume objects in the workspace.

To learn more about this app, see Get Started with Medical Image Labeler.

Note

The app is not supported in MATLAB^® Online™. For details, see Specifications and Limitations.

×

Open the Medical Image Labeler App

• MATLAB Toolstrip: On the Apps tab, under Image Processing and Computer Vision, click the Medical Image Labeler app icon.

• MATLAB command prompt: Enter medicalImageLabeler.

Examples

• Get Started with Medical Image Labeler
• Visualize 3-D Medical Image Data Using Medical Image Labeler
• Label 2-D Ultrasound Series Using Medical Image Labeler
• Label 3-D Medical Image Using Medical Image Labeler
• Automate Labeling in Medical Image Labeler
• Get Started with MONAI Label in Medical Image Labeler
• Segment CT Scan Using MONAI Label
• Collaborate on Multi-Labeler Medical Image Labeling Projects

Programmatic Use

medicalImageLabeler

medicalImageLabeler opens the Medical Image Labeler app, which enables you to start a new session to label 2-D or 3-D medical image data.

medicalImageLabeler(gTruthFile)

medicalImageLabeler(gTruthFile) opens the app and loads the image and label data stored in the file gTruthFile into the app. gTruthFile is the full path to a MAT file containing a groundTruthMedical object, specified as a string scalar or character vector.

medicalImageLabeler(gTruth)

medicalImageLabeler(gTruth) opens the app and loads the image and label data stored in the groundTruthMedical object gTruth from the workspace into the app.

medicalImageLabeler(sessionFolder)

medicalImageLabeler(sessionFolder) opens the app and loads a saved labeling session into the app, where sessionFolder is the full path to a session folder created using the Medical Image Labeler app.

medicalImageLabeler(sessionType)

medicalImageLabeler(sessionType) opens the app and creates a new session of the specified type, where sessionType is "Volume" or "Image". For more information about session types, see Get Started with Medical Image Labeler.

Version History

Introduced in R2022b

R2024b: Import medicalVolume object from workspace, new file formats

• The app is more flexible with regard to the data formats you can import.

  • In an image session, you can now import data from all graphics file formats listed in the imformats image file format registry, or import medicalImage objects from the MATLAB workspace. This enhancement provides more flexibility in the types of 2-D data you can label, including data with less complete metadata than medical file formats like DICOM. When you import a file in a format other than DICOM, the app assumes default pixel spacing of 1-by-1 millimeters.

  • In a volume session, you can now import data from TIFF files, or import medicalVolume objects from the MATLAB workspace. This enhancement provides more flexibility in the types of 3-D data you can label with the app, including data with less complete metadata than medical file formats like DICOM and NIfTI. To account for limited spatial information, the app makes these changes when you load data from TIFF files or medicalVolume objects with incomplete spatial referencing:

    • Assumes default spatial referencing with voxel spacing of 1-by-1-by-1 millimeters, and the first voxel located at (1, 1, 1).

    • Hides anatomical plane and direction labels, and labels 2-D slice panes as Direction 1, Direction 2, and Direction 3.

    • Saves label data as a MAT file containing a medicalVolume object with default spatial referencing that matches the labeled image data.

    The app behavior has not changed for data loaded from DICOM, NIfTI, and NRRD files.

R2024b: Improved performance when importing large data sets

The app has faster performance when importing many data files into a labeling session. The performance improvement is more noticeable if you import larger numbers of files. The performance improvement is similar for 2-D and 3-D data.

For example, if you import 800 X-ray images from a groundTruthMedical object, performance is approximately 10x faster in R2024b than in R2024a. The approximate wait times during import are 17.5 s in R2024b and 184.2 s in R2024a. The times were measured on a Windows^® 11 Intel^® Xeon^® W-2133 CPU @ 3.60GHz test system.

As a result of the enhancement, the Data Browser pane has a different appearance. The new Data Browser has labels for the columns indicating the names of the data files Data Name and label status of each file Has Labels. The Has Labels column uses a green check mark to indicate that a file contains labels. In prior releases, data files with no labels were indicated by a no labels symbol.

R2024a: MONAI Label deep learning models, new preferences for superpixels edge color and autosave

• Connect to the Medical Open Network for AI (MONAI) Label platform from within the Medical Image Labeler app to apply fully automated and interactive deep learning models for segmenting radiology images. This functionality requires the Medical Imaging Toolbox Interface for MONAI Label Library support package. You can install the support package from Add-On Explorer. For more information about installing add-ons, see Get and Manage Add-Ons. The support package also requires Deep Learning Toolbox™ and Computer Vision Toolbox™.

• Change the edge color for the Paint by Superpixels tool to increase the contrast between the superpixel boundaries and the image to label. To change the edge color, go to the Labeler tab and select Preferences. Under Paint by Superpixels, select the Superpixel Edge Color color box, and select a new color from the color picker.

• You can now disable autosave. By default, the app periodically pauses to save your ground truth data, which can take several seconds for very large image files. Disabling autosave avoids pauses, but can lead to accidental loss of data. To turn off autosave, in the Labeler tab, select Save Session and clear the Auto Save checkbox.

• Easily disable drawing tools when you want to explore an image without modifying labels. Starting this release, clicking the toolstrip button for an enabled drawing tool, such as Paint Brush or Trace Boundary, disables the tool. Click the button again to enable the tool and continue drawing.

R2023b: Specify window and level, set display preferences

• Adjust image brightness and contrast by typing values in the Window Level and Window Width boxes in the app toolstrip. In prior releases, the toolstrip displayed Window Bounds values, but these boxes were not editable.

• Use the new Preferences dialog box to set app preferences. Set the Display Convention preference to view volume data using the Radiological or Neurological display convention. Set the Window Level preference to control the mouse action shortcuts for adjusting image brightness and contrast.

R2023a: Generate animations and display slice crosshairs

The app includes these new capabilities:

• Animation Generator — Configure, preview, and export animations from the app. In a volume session, generate animations that loop through slices in the coronal, sagittal, or transverse plane, or show a rotating 3-D volume. In an image session, generate animations that loop through the frames of an image series, such as an ultrasound video.

  Optionally include labeled regions in animations, and customize parameters such as step size, loop count, and total animation length in seconds. To open the Animation pane, in the app toolstrip, click Generate Animations.

• Crosshair Navigation — In a volume session, view and navigate 2-D slice positions using crosshairs. To turn on the crosshair indicators, in the app toolstrip, select Crosshair Navigation. The indicators show the relative positions of the slices in the other 2-D views. To navigate slices, pause on a crosshair until the cursor changes to the fleur shape, , and then click and drag to a new position. The other slice views update automatically. To hide the crosshairs, in the app toolstrip, clear Crosshair Navigation.