Abstract:

A new way for extracting sound from images is described using the bidimensional Fast Fourier Transform (FFT) of digitalized microscopic images of human cells. The FFT images provide information on the presence of cyclic harmonic structures inherent in the texture of the original image, since each pixel of the FFT image defines a frequency and an amplitude. Reading the FFT image by a pointer rotating in a clockwise manner, it is possible to transform this information into a sound equivalent to the original image. This method is illustrated with images from cells of benign and malignant human neoplastic tissue, which can be differentiated by their equivalent "cell sound".