Speckle produced by strongly-scattering media contains information about its optical properties. Statistical speckle study allows discrimination between media and enables one to characterize any change. Two approaches of the speckle phenomenon are used in the measurement of speckle produced by monodisperse-polystyrene microspheres in solution and mixtures of them: a stochastic approach based on the fractional Brownian motion and a classical frequential approach based on speckle size measurement. In this paper, we introduce an approach that contains the multi-scale aspect of the speckle; therefore it provides more information on the medium than the speckle dimension. The obtained results show that the stochastic approach allows a better samples discrimination than the classical frequential approach.

Abstract. Radiation burns account for the vast majority of damage by accidental radiation exposure. They are characterized by successive and unpredictable inflammatory bursts that are preceded by a clinically latent postirradiation period. Diagnosis and prognosis of the clinical course of radiation burns have proven to be a difficult task. In a classical clinical setting, no technique can distinguish irradiated versus healthy skin during the clinically latent period, hence development of new tools is required. This work describes a noninvasive technique based on speckle phenomenon, designed to support radiation burn diagnosis and prognosis. Speckle produced by strongly scattering media contains information about their optical properties. The difficulty is to extract significant information from speckle patterns to discriminate between strongly scattering media and to characterize any change. Speckle patterns from irradiated and nonirradiated porcine skins are recorded in vivo several times after radiation exposure. A fractal approach is used in the treatment of speckle patterns. The results show that this technique allows discrimination between healthy and irradiated skin, in particular during the clinically latent period ͑p Ͻ 0.01͒. Parameters extracted from speckle patterns discriminate and vary differently with radiation, which means they represent different information about skin changes.

In this paper, we present a new approach of the speckle phenomenon. This method is based on the fractal Brownian motion theory and allows the extraction of three stochastic parameters to characterize the speckle pattern. For the first time, we present the results of this method applied to the discrimination of the healthy vs. pathologic skin. We also demonstrate, in case of the scleroderma, than this method is more accurate than the classical frequential approach.