3-D Volumetric Image Processing

3-D volumetric images, or volumes, represent the brightness or color of a scene sampled in three spatial dimensions: height, width, and depth. Volumes are used in a variety of fields, such as:

Medical imaging, which analyzes structures in the body to detect irregularities and measure tissue properties.
Manufacturing, which examines goods for quality and improves product design.
Geological imaging, which performs nondestructive testing of the environment and examines properties of rocks and fossils.
Archaeology and art restoration, which digitally preserve artifacts and conduct non-invasive analysis of archaeological findings.

Image Processing Toolbox™ represents volumes as multidimensional arrays with three spatial dimensions. Grayscale and binary volumes are typically 3-D arrays, where each (row, column, plane) index of the array corresponds to a single voxel in the volume. Color volumes are typically 4-D arrays, where the fourth dimension stores the intensity values of the color channels.

Note

Image Processing Toolbox also uses multidimensional arrays to represent image sequences. Unlike volumes, image sequences typically have a temporal dimension that stores samples of the scene over time. Some functions work for both volumetric images and image sequences, whereas other functions work for only one type of input. For more information, see Work with Image Sequences as Multidimensional Arrays.

This page provides a list of functions that support 3-D volumetric images in Image Processing Toolbox.

Volume Import and Export

Function	Description
Blocked Volumes
`blockedImage`	Big or multiresolution image made from smaller discrete blocks
`blockedImageDatastore`	Datastore that reads blocks from `blockedImage` objects
`makeMultiLevel3D`	Create 3-D multilevel blocked image with different resolutions from a single level of a blocked image
Standard File Formats
`dicomread`	Read DICOM image
`dicomreadVolume`	Create 4-D volume from set of DICOM images
`dicomContours`	Extract ROI data from DICOM-RT structure set
`niftiinfo`	Read metadata from NIfTI file
`niftiwrite`	Write volume to file using NIfTI format
`niftiread`	Read NIfTI image
`tiffreadVolume`	Read volume from TIFF file
`medicalVolume` (Medical Imaging Toolbox)	3-D medical image voxel data and spatial referencing information (requires Medical Imaging Toolbox™)

Display

Function	Description
Interactively Display and Explore Volumes
Volume Viewer	View volumetric data and labeled volumetric data, and interactively adjust display settings
Render Volumes in 3-D Space
`viewer3d`	Create 3-D viewer with scene-level controls
`volshow`	Display volume in 3-D viewer
`Surface`	Display surface in 3-D viewer
Display 2-D Cross Sections (Slices)
`montage`	Display multiple image slices as rectangular montage
`sliceViewer`	Browse image slices in grayscale or RGB volume
`orthosliceViewer`	Browse orthogonal slices in grayscale or RGB volume
`obliqueslice`	Extract oblique slice from volume

See these related examples:

Geometric Transformations and Registration

Function	Description
Resize, Rotate, and Crop Volumes
`imresize3`	Resize 3-D grayscale volume
`imrotate3`	Rotate 3-D grayscale volume
`imcrop3`	Crop 3-D volume
Create and Apply 3-D Geometric Transformation
`imwarp`	Apply geometric transformation
`transltform3d`	3-D translation geometric transformation
`rigidtform3d`	3-D rigid geometric transformation
`simtform3d`	3-D similarity geometric transformation
`affinetform3d`	3-D affine geometric transformation
`randomAffine3d`	Randomized 3-D affine geometric transformation
`transformPointsForward`	Apply forward geometric transformation
`transformPointsInverse`	Apply inverse geometric transformation
Register 3-D Volumes
`imregister`	Intensity-based volume registration
`imregdemons`	Estimate displacement field that aligns two volumes
Medical Registration Estimator (Medical Imaging Toolbox)	Interactively register volumes using the Medical Registration Estimator app (requires Medical Imaging Toolbox)
Store 3-D Spatial Referencing
`imref3d`	Reference volume to world coordinates
`affineOutputView`	Spatial extent of transformed volume

For an example, see Register Multimodal 3-D Medical Images.

Filtering and Enhancement

Function	Description
Apply Filter to Volume
`imfilter`	Apply N-D filter to multidimensional volumes
`fspecial3`	Create predefined 3-D filter
`medfilt3`	3-D median filtering
`imgaussfilt3`	3-D Gaussian filtering
`imboxfilt3`	3-D box filtering
`imnoise`	Add noise to volume
`integralBoxFilter3`	Box filtering of 3-D integral volumes
`integralImage3`	3-D integral volume
`fibermetric`	Enhance elongated or tubular structures in volume using Frangi vesselness filter
`padarray`	Pad array
Adjust Contrast
`histeq`	Enhance contrast using histogram equalization
`imadjustn`	Adjust intensity values
`imhistmatchn`	Adjust histogram of volume to match reference histogram
Deblur Volumes
`deconvblind`	Deblur volume using blind deconvolution
`deconvlucy`	Deblur volume using Lucy-Richardson method
`deconvreg`	Deblur volume using regularized filter
`deconvwnr`	Deblur volume using Wiener filter
`edgetaper`	Taper discontinuities along volume edges
`otf2psf`	Convert optical transfer function to point-spread function
`psf2otf`	Convert point-spread function to optical transfer function

Segmentation

For more information about image and volume segmentation techniques, see Get Started with Image Segmentation.

Function	Description
Segment Volume Interactively
Volume Segmenter	Interactively segment objects in a volume and refine segmentation masks using the Volume Segmenter app
Medical Image Labeler (Medical Imaging Toolbox)	Interactively label volumes using manual, semiautomated, and automated techniques including deep learning networks (requires Medical Imaging Toolbox)
Binary Volume Segmentation
`activecontour`	Segment grayscale volume into foreground and background using active contours (snakes) region growing technique
`lazysnapping`	Segment grayscale volume into foreground and background using graph-based segmentation
`grabcut`	Segment grayscale volume into foreground and background using iterative graph-based segmentation
`adaptthresh`	Adaptive volume threshold using local first-order statistics
`graythresh`	Global volume threshold using Otsu's method
`gradientweight`	Calculate weights for voxels based on volume gradient
`graydiffweight`	Calculate weights for image pixels based on grayscale intensity difference
`grayconnected`	Select contiguous volumetric region with similar gray values using flood-fill technique
`imbinarize`	Binarize volume by thresholding
`imsegfmm`	Binary volume segmentation using fast marching method
Segment Volumes into Labeled Volumes
`imsegkmeans3`	K-means clustering based volume segmentation
`superpixels3`	3-D superpixel oversegmentation of volume
`watershed`	Watershed transform
Evaluate Segmentation Results
`bfscore`	Contour matching score
`dice`	Sørensen-Dice similarity coefficient
`jaccard`	Jaccard similarity coefficient

See these related examples:

Morphology

To ensure a volume is processed as a volume and not an image sequence, use a 3-D structuring element.

Function	Description
Morphological Operations
`bwmorph3`	Morphological operations on binary volume
`imclose`	Morphologically close volume
`imdilate`	Dilate volume
`imerode`	Erode volume
`imopen`	Morphologically open volume
`imbothat`	Bottom-hat filtering
`imtophat`	Top-hat filtering
`bwskel`	Reduce all objects to lines in binary volume
`bwhitmiss`	Binary hit-miss operation
`bwperim`	Find perimeter of objects in binary volume
`imclearborder`	Suppress light structures connected to volume border
`imkeepborder`	Retain light structures connected to volume border
`bwulterode`	Ultimate erosion
`imfill`	Fill regions and holes in volume
`padarray`	Pad volume
Morphological Reconstruction
`imreconstruct`	Morphological reconstruction
`imregionalmax`	Regional maxima
`imregionalmin`	Regional minima
`imextendedmax`	Extended-maxima transform
`imextendedmin`	Extended-minima transform
`imhmax`	Suppress regional maxima using H-maxima transform
`imhmin`	Suppress regional minima using H-minima transform
`imimposemin`	Impose minima
Structuring Elements
`strel`	Morphological structuring element for binary volumes
`offsetstrel`	Morphological offset structuring element for grayscale volumes
`conndef`	Create connectivity array

Image and Object Analysis

Function	Description
Analyze Objects
`regionprops3`	Measure properties of 3-D regions
`bwselect3`	Select objects in binary volume
`bwconncomp`	Find and count connected components in binary volume
`bwareaopen`	Remove small objects from binary volume
Evaluate Image Quality
`immse`	Mean-squared error
`psnr`	Peak signal-to-noise ratio (PSNR)
`ssim`	Structural similarity (SSIM) index
`multissim3`	Multiscale structural similarity (MS-SSIM) index
Measure Image Properties
`imhist`	Histogram of volume data
`bwdist`	Distance transform of binary volume
`bwdistgeodesic`	Geodesic distance transform of binary volume
`graydist`	Gray-weighted distance transform
Detect Edges
`edge3`	Find edges in grayscale volume
`imgradient3`	Find gradient magnitude and direction of volume
`imgradientxyz`	Find directional gradients of volume
Measure Texture
`entropy`	Entropy of grayscale volume
`entropyfilt`	Local entropy of grayscale volume
`rangefilt`	Local range of volume
`stdfilt`	Local standard deviation of volume

Deep Learning with Volumes

For more information, see Get Started with Image Preprocessing and Augmentation for Deep Learning (Computer Vision Toolbox) and Preprocess Volumes for Deep Learning (Deep Learning Toolbox).

Function	Description
Read and Preprocess Data for Deep Learning
`blockedImageDatastore`	Datastore that reads blocks from `blockedImage` objects
`randomPatchExtractionDatastore`	Datastore that extracts random 3-D patches from volumes or pixel label volumes
`randomAffine3d`	Randomized 3-D affine transformation
`centerCropWindow3d`	Centered cuboidal cropping window
`randomCropWindow3d`	Randomized cuboidal cropping window
Create Deep Neural Networks
`resize3dLayer`	Resize 3-D layer in neural network (requires Deep Learning Toolbox™)
`dlresize`	Resize spatial dimensions of `dlarray` object (requires Deep Learning Toolbox)

For an example, see 3-D Brain Tumor Segmentation Using Deep Learning.

Arithmetic

Function	Description
`imabsdiff`	Absolute difference of two volumes
`imadd`	Add two volumes or add constant to volume
`imdivide`	Divide one volume into another or divide volume by constant
`immultiply`	Multiply two volumes or multiply volume by constant
`imsubtract`	Subtract one volume from another or subtract constant from volume