Calculating affine transformation using corresponding points

Hello, I have a question about feature-based image registration process. I have two 3D images (volumes) with a set of N predefined corresponding points of [x,y,z,X,Y,Z] (where [x,y,z] and [X,Y,Z] are the coordinates of corresponding points in reference/imgR and target/imgT images, correspondingly). I am trying to align them based on these corresponding points using affine transformation.

Here are the steps I have gone through to register imgR and imgT volumes using 4 points:

• Four sets of corresponding points are selected randomly.
• Affine transformation matrix (T) is calculated using selected corresponding points (see below).

• Target image is resampled based on the calculated T using MATLAB's built-in imwarp function.

This works okay but it is not so accurate. How can I use all of the corresponding points instead of 4 of them?