Input Files

Movies

Movies should be stored as image sequences, image stacks, or in the Nikon ND2 format.

Image Sequences (Individual Images)

Each file contains an image from one channel at one time point. All files from the same channel should locate in the same folder.

Image Stacks (Multi-Page Images)

TIFF or GIF formats only. Each stack contains the images from one channel at all time points.

Nikon ND2 Format

A movie can be stored in multiple segments (files). Each segment contains the images from all channels at all time points within the segment. This requirement is usually satisfied if the movie is captured by the Nikon NIS Element software.

The first channel should image cell nuclei. Images of this channel will be used for Segmentation and Track Linking. Images of the other channels will be used for Signal Extraction.

Camera Dark Noises (Optional)

Camera dark noises refer to the pixel intensities without illumination. Also known as CMOS Offset.

Camera dark noises are represented in a MATLAB data file (.mat) with a single matrix called cmosoffset. cmosoffset has the same dimension as the movie images. Each element stores the value of camera dark noises at the corresponding pixel.

Procedure for Generating cmosoffset

• Capture a few images of an empty well with the laser switched off.

• Load the images into MATLAB as a cell array called all_images where all_images{i} stores the i-th captured image.

• Execute the following scripts

  cmosoffset = zeros(size(all_images{1})); % initialize the matrix
  for i=1:length(all_images) % adding the intensity values of each image
      cmosoffset = cmosoffset + double(all_images{i});
  end
  cmosoffset = cmosoffset / length(all_images); % average the intensities
  save('cmosoffset.mat', 'cmosoffset'); % save cmosoffset
  

Illumination Bias (Optional)

Illumination bias refers to the uneven laser illumination within the field of view. Correction of this bias is also known as shading correction.

Illumination bias is fluorescence channel-dependent. For each channel, it is represented in a MATLAB data file (.mat) with a single matrix called bias. bias has the same dimension as the movie images. Each element stores the relative intensity at the corresponding pixel compared to the image average.

Procedure for Generating bias

• Capture a few images of a well with medium in the channel of interest.

• Load the images into MATLAB as a cell array called all_images where all_images{i} stores the i-th captured image.

• Execute the following scripts

  bias = zeros(size(all_images{1})); % initialize the matrix
  for i=1:length(all_images) % compute the bias from each image
      temp = double(all_images{i}) - cmosoffset; % subtract camera dark noise
      temp = temp / mean(temp(:)); % normalize the image
      bias = bias + temp;
  end
  bias = bias / length(all_images); % average the bias
  save('bias.mat', 'bias'); % save the bias
  

Before analysis, EllipTrack removes the camera dark noises and illumination bias from each image.

I = (double(I) - cmosoffset) ./ bias;

Here, I refers to an image and ./ refers to elementwise-division.