Display, Segment, and Process Medical Imaging Data with MATLAB - MATLAB
Video Player is loading.
Current Time 0:00
Duration 4:30
Loaded: 3.67%
Stream Type LIVE
Remaining Time 4:30
 
1x
  • Chapters
  • descriptions off, selected
  • en (Main), selected
    Video length is 4:30

    Display, Segment, and Process Medical Imaging Data with MATLAB

    The Medical Image Labeler app, released with the new Medical Imaging Toolbox™, is designed to visualize, segment, and process medical images in MATLAB®.

    Import CT scans, MRI, ultrasound, or microscopy medical imaging data directly into the app from DICOM, NIfTI, or NRRD formatted files. Analyze the data by displaying the images slice by slice in the transverse, sagittal, and coronal planes of view. Visualize the entire volume in the 3D display panel with your preferred rendering options.

    Segment and label the images and volumes manually or using one of the many automated techniques featured in the app. Use “Active Contours,” “Paint by Super Pixels,” “Edge Smoothing,” and many more functionalities to speed up the labeling workflow. Import your image processing or deep learning algorithms into the app for customized segmentation procedures.

    Use the app to display the final results and publish them in PDF format. Export the labeled volumes and use them for diagnostic or radiomics applications.

    Published: 6 Feb 2023

    In this video, we will highlight the main features of the Medical Image Labeler App. This app, released with the new Medical Imaging Toolbox, is specifically designed to help you visualize, segment, and process Medical Imaging Data in MATLAB. The first step of the workflow is to import and visualize medical data. The app supports the most common medical imaging formats, such as DICOM, NIfTI, and NRRD. The metadata from the files is used to display the objects correctly oriented and with the appropriate anatomical markers in the transverse, sagittal, and coronal points of view.

    Volumes are displayed with the correct spatial referencing in the 3D Volume Display panel, which provides a smooth and interactive visualization of medical data in 3D. You can use linear grayscale to highlight details at different depths using varying opacity levels or you can use one of the preset modes to focus on specific parts of the body, like the bones, the lungs, the soft tissues, or the coronary blood vessels. Additionally, you have the option to create customized 3D renderings. Here, for example, I designed one that displays both the bones and the lungs of the patient.

    The Medical Image Labeler App makes segmenting medical data much easier. Segmentation is used for identifying objects of interest in images and volumes and it's a critical part of the radiomics and deep-learning workflows. The app features semi-automated segmentation techniques, like the system-freehand Mode, which leads your cursor by following the contours of objects in the image. The Paint by Super Pixels mode uses clustering to separate the data into regions and then lets you paint over the ones that you want to select. With the new Medical Imaging Toolbox, we also introduced a brand new technique, the Trace Boundary Mode, which can automatically detect boundaries of objects.

    For 3D volumes, such as a stack of CT slices, you also have the option to semi-automate their segmentation process. Starting from just a few sample slices, the app will propagate the segmented regions throughout the entire volume. Depp also offers a variety of fully automated segmentation algorithms, which are applied either slice by slice or to entire volume at once. These include active counters, a region-growing segmentation technique, which expands existing segmented regions into current and in the neighboring slices. And edge smoothing, which uses morphological operations to refine the segmented regions.

    If the built-in techniques are insufficient, there is also the option to use custom segmentation algorithms. Whether it's a complex image processing technique or a pre-trained segmentation neural network, it can easily be imported as a function and used to segment the data directly inside of the app. Here, I'm showing how a pre-trained neural network is used to identify the lungs inside the medical volume. The segmented regions can then be visualized interactively in the 3D Volume Viewer using your preferred rendering option. You can also make customized visualizations that can be published in PDF format ready to be shared with colleagues or incorporated in medical articles.

    Once I finish segmenting the data, with the press of a button, you can export the groundtruth data and the segmented images and volumes to the MATLAB workspace or to a local file. The groundtruth data is ready to be used to train a neural network to perform detection, classification, or even segmentation of medical data. The segmented images and volumes can be used to do some radiomic analysis, like measuring physical properties, such as the volume of the lungs. You can also display the segmented data either directly inside of the app or using one of the many MATLAB tools, introduce clipping planes to visualize the volume at specific depth or display the medical volume as sliced planes.

    I hope you enjoyed this video that showcase the main workflows of MATLAB's new Medical Image Labeler App. Now go try it out for yourself. Thank you for listening. And if you want more information and resources, please visit the medical Imaging Toolbox product page.

    View more related videos