Segmentation

Medical image segmentation using deep learning, interactive labeling app, or image processing algorithms

Image segmentation is the process of partitioning an image into regions. Semantic segmentation associates each pixel or voxel in an image with a class label that describes the meaning of an image region, such as bone, tumor, or background. You can perform medical image semantic segmentation using deep learning, the interactive Medical Image Labeler app, or image processing algorithms. Deep learning workflows require Deep Learning Toolbox™ and Computer Vision Toolbox™.

Apps

Medical Image Labeler

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

Functions

expand all

Segmentation Algorithms

`imbinarize`	Binarize 2-D grayscale image or 3-D volume by thresholding
`graythresh`	Global image threshold using Otsu's method
`multithresh`	Multilevel image thresholds using Otsu’s method
`otsuthresh`	Global histogram threshold using Otsu's method
`adaptthresh`	Adaptive image threshold using local first-order statistics
`grayconnected`	Select contiguous image region with similar gray values using flood-fill technique
`watershed`	Watershed transform
`activecontour`	Segment image into foreground and background using active contours (snakes) region growing technique
`imsegfmm`	Binary image segmentation using fast marching method
`gradientweight`	Calculate weights for image pixels based on image gradient
`graydiffweight`	Calculate weights for image pixels based on grayscale intensity difference
`imsegkmeans`	K-means clustering based image segmentation
`imsegkmeans3`	K-means clustering based volume segmentation
`superpixels`	2-D superpixel oversegmentation of images
`superpixels3`	3-D superpixel oversegmentation of 3-D image

Deep Learning

Load and Prepare Training Data

`groundTruthMedical`	Ground truth label data for medical images
`changeFilePaths`	Change file paths in ground truth data for medical images
`merge`	Merge two or more `groundTruthMedical` objects
`imageDatastore`	Datastore for image data
`pixelLabelDatastore`	Datastore for pixel label data
`randomPatchExtractionDatastore`	Datastore for extracting random 2-D or 3-D random patches from images or pixel label images
`combine`	Combine data from multiple datastores
`transform`	Transform datastore

Import Network

`importKerasLayers`	Import layers from Keras network
`importTensorFlowLayers`	Import layers from TensorFlow network
`importONNXLayers`	Import layers from ONNX network

Design Networks

`fcnLayers`	Create fully convolutional network layers for semantic segmentation
`segnetLayers`	Create SegNet layers for semantic segmentation
`unet3dLayers`	Create 3-D U-Net layers for semantic segmentation of volumetric images
`unetLayers`	Create U-Net layers for semantic segmentation
`pixelClassificationLayer`	Create pixel classification layer for semantic segmentation
`dicePixelClassificationLayer`	Create pixel classification layer using generalized Dice loss for semantic segmentation

Segment Images

semanticseg Semantic image segmentation using deep learning

Visualize Segmentation Results

`labeloverlay`	Overlay label matrix regions on 2-D image
`volshow`	Display volume

Evaluate Segmentation Accuracy

`jaccard`	Jaccard similarity coefficient for image segmentation
`dice`	Sørensen-Dice similarity coefficient for image segmentation
`bfscore`	Contour matching score for image segmentation

Topics

Get Started with Image Preprocessing and Augmentation for Deep Learning
Preprocess data with deterministic operations such as normalization or color space conversion, or augment training data with randomized operations such as random cropping or color jitter.
Create Datastores for Medical Image Semantic Segmentation

Create datastores that contain images and pixel label data from a groundTruthMedical object for training semantic segmentation deep learning networks.
- Convert Ultrasound Image Series into Training Data for 2-D Semantic Segmentation Network
- Create Training Data for 3-D Medical Image Semantic Segmentation
Datastores for Deep Learning (Deep Learning Toolbox)
Learn how to use datastores in deep learning applications.
List of Deep Learning Layers (Deep Learning Toolbox)
Discover all the deep learning layers in MATLAB^®.

Related Information

Image Segmentation

Featured Examples

Segment Lungs from CT Scan Using Pretrained Neural Network

Import a pretrained ONNX™ (Open Neural Network Exchange) 3-D U-Net [1] and use it to perform semantic segmentation of the left and right lungs from a 3-D chest CT scan. Semantic segmentation associates each voxel in a 3-D image with a class label. In this example, you classify each voxel in a test data set as belonging to the left lung or right lung. For more information about semantic segmentation, see Semantic Segmentation.

Open Live Script