Data Format

A collection of T images acquired under a single experimental condition is represented by T x P  matrix M, which is an image stack,
 

$$\mathbf{M}=\left[ \begin{array}{c}f(1,\mathbf{x})\\f(2,\mathbf{x})\\\vdots \\f(T,\mathbf{x})\end{array}\right]$$

where each row is an image vector f(t,x) and x = 1,...,P is the index to the pixel position. It does not matter how an image is converted into a P-dimensional vector, but you need to be able to undo this conversion in order to display it as an image. You may want to use our img2vec.m and vec2img.m routine to go back and forth between the vector and the image representations.

The full data matrix F  is constructed by stacking these M matrices for each of K experimental conditions,
 

$$\mathbf{F}=\left[ \begin{array}{c}\mathbf{M}_{1}\\\mathbf{M}_{2}\\\vdots \\\mathbf{M}_{K}\end{array}\right]$$

where M_k is the image stack under condition k. Hence, F  is a KT x P matrix. Various pre-processing may (optionally) be applied to the data, such as spatial/temporal filters, subtraction/division by control images, etc. A spatial filtering routine, Gconv2.m, (2-dimensional convolution with circularly symmetric Gaussian kernel) is included in this package, and may be directly applied to the data matrix F.