nvis is a matrix viewer for matrices of any size. Its aimed towards MATLAB users that need to asses a lot of higher-dimensional data in their daily workflow. A 2-dimensional slice of the matrix is shown, and higher
Multidimensional: Visualization should work with matrices independent of their dimensionality. While the display is limited to two dimensions at a time, selecting datasets along the remaining dimensions (e.g. echo, channel, time) should be quick and intuitive. Many available tools are limited to three-dimensional matrices.
Versatile: For the development of sequences and especially image reconstruction routines, visual feedback and the ability to quickly assess images, k-spaces, raw-data, sensitivity maps or filters are key. Complex values must be supported, as well as data type such as int or bool; the content of an matrix should not be relevant.
Dynamic: Dynamic data is best visualized in motion, while still allowing for easy comparison along different dimensions (e.g. channels).
Comparative: Overlaying two images is a simple and intuitive way of comparing two reconstructions form the same data, prediction and ground truth, or the alignment of a ROI to the image.
Interactive: Windowing allows for intuitive selection of image contrast in image space as well as k-space. Images with the current contrast can be saved from the GUI or using only command line arguments. This is also true for saving videos or animations.
Quantitative: We don’t need the next fully featured image analysis tool, but extracting pixel values and drawing signal and noise ROIs should be supported.
Lightweight: matrices containing medical images can require a lot of memory, especially when 3D, multi-channel or functional information are included. The requirements increase further, when data is stored as high-precision complex values. Thus, no additional memory should be required when visualizing the data.
nvis can be called with one
nvis(A)
or two input matrices
nvis(A, B)
B must have the same size as A. Except: A and B can differ in size along one or more dimensions, as long as either A or B are singleton. E.g.
nvis(rand(100, 100, 10, 1), rand(100, 100, 1, 5))
For dimensions where all input matrices are singleton, no slider is shown.
The limits of the colormap can be changed with center and width values. These can also be varied interactively by pressing middle or right mouse button on the canvas and dragging the mouse.
Matrix dimensions can be labelled:
nvis(rand(100, 100, 10), 'dimLabel', {'dx', 'dy', 'time'}
Values along each dimension can be labelled with any string. When providing an empty matrix [], the values are enumerated, e.g. 1:size(..., 1).
nvis(rand(100, 100, 10), 'dimVal', {[],[], 0.1*(1:10)})
For complex data, either the absolute value, phase, real- or imaginary part can be displayed. It is also possible to show the 2D Fourier transform of the currently selected slice.
The shown image/animation can also be saved to file. Either from the GUI, or - without opening the GUI -- from the command line.
nvis(imageMat,'CW',[0.5,1],'Colormap','parula','SaveImage','image.png')
nvis(imageMat,'CW',[0.5,1],'Colormap',parula(32),'fps',60,'SaveVideo','animation.gif')
By default, the matrix is displayed as a square. To set the aspect ratio according to the matrix size, use
nvis(A, 'aspectRatio', 'image')
More instructions are documented under
help nvis