Loading Microscopy Volumes

Image volumes created by advanced micrography, such as confocal imaging, are best loaded into MATLAB using the Bio-formats MATLAB toolbox, bfmatlab. This toolbox can handle the proprietary file formats generated by all of the major microscope manufacturers. Saving image stacks in the proprietary file format of the microscope best ensures that the full resolution and bit-depth of the image data is preserved, along with all captured metadata.

In this example, we will load a 4D image stack of a taste bud.

This image stack has 4 Dimensions. Its size is 512x512x3x12, which means that each individual slice in the stack is 512x512 and has three channels (like an RGB image). And there are 12 of these slices.

mmReadImgND

The course function mmReadImgND is a wrapper for the tools included in the bfmatlab toolbox. The following syntax loads a color image stack created by a confocal microscope.

[RGB4,meta] = mmReadImgND("2641-tastebud-RGB.tif");

Reviewing the workspace shows the size and class of RGB4…

result

  Name        Size                    Bytes  Class     Attributes

  RGB4      512x512x3x12            9437184  uint8               
  meta        1x1                      3775  struct              

…RGB4 is an 8-bit 4D array. It has 3 channels and 12 image slices. Each image slice in the volume has 512 rows (Height) and 512 columns (Width).

Reviewing the meta structure reveals the same information , along with size of the voxels (PixelSpacing) in this volume:

meta = struct with fields:
            filename: "2641-tastebud-RGB.tif"
    StudyDescription: '2641-tastebud-RGB.tif'
      SeriesSelected: 0
               Width: 512
              Height: 512
            BitDepth: 'uint8'
        PixelSpacing: [0.30333 0.30333 0.75531]
        ChannelCount: 3
          PlaneCount: 12
           TimeCount: 1
                 XML: "<?xml version="1.0" encoding="utf-8"?>↵<OME xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openmicroscopy.org/Schemas/OME/2016-06 http://www.openmicroscopy.org/Schemas/OME/2016-06/ome.xsd">↵<Image ID="Image:0" Name="2641-tastebud-RGB.tif">↵<Description/>↵<Pixels BigEndian="true" DimensionOrder="XYCZT" ID="Pixels:0" Interleaved="false" PhysicalSizeX="0.303326643969745" PhysicalSizeXUnit="µm" PhysicalSizeY="0.303326643969745" PhysicalSizeYUnit="µm" PhysicalSizeZ="0.755310778914241" PhysicalSizeZUnit="µm" SignificantBits="8" SizeC="3" SizeT="1" SizeX="512" SizeY="512" SizeZ="12" Type="uint8">↵<Channel ID="Channel:0:0" SamplesPerPixel="1">↵<LightPath/>↵</Channel>↵<Channel ID="Channel:0:1" SamplesPerPixel="1">↵<LightPath/>↵</Channel>↵<Channel ID="Channel:0:2" SamplesPerPixel="1">↵<LightPath/>↵</Channel>↵<MetadataOnly/>↵</Pixels>↵</Image>↵</OME>"

Using imread

The imread function can be used to image stacks saved in the .TIF file format.

To load an image stack, you first need information about that stack. In this example, we will use this .TIF file

The function imfinfo returns the metadata for an image stack.

meta = imfinfo("TasteBud3D.tif")

Output from imfinfo

meta = 

  12×1 struct array with fields:

    Filename
    FileModDate
    FileSize
    Format
    FormatVersion
    Width
    Height
    ⋮

…For this example file, we get a lot of metadata (more than 39 fields). Notice that meta is a 12X1 structure, so in fact, you get a metadata structure element for each slice in the image stack. This is one way to determine the number of slices in the image stack. Most of the data is repeated for each slice, so we can examine the metadata for the first image slice to get information about the image slices:

Review element 1 of the metadata structure

numel(meta) % number of slices

ans = 

    12

meta(1).Width % number of columns

ans =

   512

meta(1).Height % number of rows

ans =

   512
meta(1).BitDepth % bit depth

ans =

     8

…Here we review the first element in the metadata structure and we find that each slice in the image stack is an unsigned 8-bit image with 512 rows (Height) and 512 columns (Width). So, the size of our image stack is 512 X 512 X 12.

The function imread can read slices from a .TIF image stack. To read in a slice, you call imread as you would for a 2D image. The following code loads the first slice from a 3D Image stack.

slice = imread("TasteBud3D.tif") % read in the first slice

If you want to load a different slice, you indicate the slice number as a second input in imread

slice3 = imread("TasteBud3D.tif",3) % read in the third slice

If you want to load the whole stack using imread, you need a FOR LOOP.

% preallocate variable
num_rows = meta(1).Height; % number of rows
num_cols = meta(1).Width; % number of columns
num_slices = numel(meta); % number of slices in the image stack
STACK = zeros(num_rows, num_cols, num_slices,'uint8'); % preallocate the variable as a uint8 3D array

% load image stack slice by slice
for n=1:num_slices
    slice = imread("TasteBud3D.tif",n); % read slice from file
    STACK(:,:,n) = slice; % add slice to incrementing location in stack
end

…When loading large datasets, it is always a good idea to preallocate the variable prior to running the FOR LOOP. Notice how we match the bit depth of the preallocated variable to the bit depth of the image stack. Failure to do so results in bit depth mismatching and much gnashing of teeth.

Reviewing the image stack in the workspace shows the following:

Output from whos

  Name               Size                    Bytes  Class     Attributes

  STACK            512x512x12              3145728  uint8               

…The image stack is loaded into MATLAB just like any other numeric array.