matlab-imagestack

Lightweight Matlas class which wraps multichannel ImageStacks

Wrapper around a 4-D tensor of image data:

Image data tensor is X by Y by channels by frames.

Properties

  .data      % stores underlying data
  .name      % descriptive string
  .frameRate % used for tvec, play, and saveVideo

Computed properties

  .imageSize - [nX nY]
  .nChannels
  .nFrames

Loading:

  s = ImageStack.fromAllInDirectory(dir, ...)
  s = ImageStack.fromTif(file, ...)

Indexing into data

  s(maskX, maskY, channels, frames) % same as s.data(...)
  s{frame} % grab single frame as matrix
  s = s.getFrames(select) % same as s.data(:, :, :, select);
  [meanVsTime, tvec] = s.vsTime() % mean over time, precede with s.crop

Return a new ImageStack:

  s = s.frames(select) % get ImageStack with selected frames
  s = s.everyNthFrame(skip) % get 1:skip:end frames
  s = s.channels(select) % get ImageStack with selected frames
  s = s.crop(selectX, selectY) % get ImageStack with selected image region

Stats:

  s.minProj
  s.maxProj
  s.meanProj
  s.globalMin
  s.globalMax
  s.globalMinMax % [min max]

Many stats functions are supported directly and act on s.data

  min(s, ...)
  max(s, ...)
  nanmin(s, ...)
  nanmax(s, ...)
  mean(s, ...)
  nanmean(s, ...)
  var(s, ...)
  nanvar(s, ...)

Standard arithmetic operators are overloaded to apply to s.data and are automatically passed through bsxfun, e.g.

  dfoverf = s ./ mean(s, 4);
  change = s - s{1};

Transformations (return image stack)

  sAlign = alignToReferenceTranslation(s, reference)
  sAlign = alignToMeanTranslation(s) % shifts each image to align with s.meanProj
  sNorm = normalize(s) % normalizes all values to [0 1] range

Visualization - uses

  s.show % imagesc on first frame
  s.play % implay on s.data
  s.imcat % output to iTerm2 terminal as an image
  s.montage(...) % montage on s.data

Chaining: any method that returns an image stack can be chained, e.g.

  load mri
  s = ImageStack(D);
  s.everyNthFrame(2).frames(1:12).normalize.montage();
  colormap gray;