Steve Eddins has developed MATLAB and image processing capabilities for MathWorks since 1993. He also coaches development teams on designing programming interfaces for engineers and scientists. Steve coauthored Digital Image Processing Using MATLAB.
Shipped with the R2022b release a couple of months ago, this product provides apps, functions, and workflows for designing and testing diagnostic imaging applications. You can perform 3D rendering and visualization, multimodal registration, and segmentation and labeling of radiology images. The toolbox also lets you train predefined deep learning networks (with Deep Learning Toolbox™). I'm looking forward to writing about this product and its capabilities.
I've been talking recently with Sailesh, one of the product developers, about the Medical Image Labeler. This app is for labeling ground truth data in 2-D and 3-D medical images. With this app, you can:
Import multiple 2-D images or 3-D image volumes.
View images as slice planes or volumes with anatomical orientation markers and scale bars.
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.
Write, import, and use your own custom automation algorithm to automatically label ground truth data.
Export the labeled ground truth data as agroundTruthMedicalobject. You can use this object to share labels with colleagues or for training semantic segmentation deep learning networks.
The Medical Image Labeler supports 2-D images and image sequences stored in the DICOM and NIfTI file formats. An image sequence is a series of images related by time, such as ultrasound data. The app supports 3-D image volume data stored in the DICOM (single or multifile volume), NIfTI, and NRRD file formats. See Get Started with Medical Image Labeler.
I asked Sailesh for his thoughts about what to tell people in this blog about Medical Image Labeler. He commented that the app is intended for people working towards any kind of AI-assisted CAD (computer-aided diagnosis). It doesn't replace a doctor or diagnostician; rather, it assists with tasks related to their research.
Sailesh also mentioned some specific capabilities that he'd like to highlight. I've prepared a 3-minute video of one of those capabilities: Using active contours to automate object labeling. This tool is one of several in Medical Image Labeler that are intended to save time in the labeling process. Check it out:
There are a few other things about the Medical Image Labeler that I hope to show you soon.
If you have suggestions for the Medical Image Labeler, or for Medical Imaging Toolbox in general, add a comment below.