“…To circumvent this problem, it is informative to note from equation (6) that most of the signal information, i.e., its energy content, lies in proximity of the least time . Thus, Medwin proposed to decompose the time signal into two parts, one having a simple form with a known exact Fourier transform, and the other component being just the left part of the di!racted "eld, and the Fourier transform of which is to be made digitally [12,13]. This latter has, of course, to be truncated somewhere in the time domain depending on the working frequency range, but mostly on the behaviour of the di!racted "eld, hence depending on the geometry of the problem.…”