A lognormal multiplicative cascade model for multifractal characterization of skin ultrasound images

Biological tissue characterization requires adequate models of the data. Research in ultrasound image modeling has mainly focused on statistical methods. Some authors have addressed the fractal properties of such images using fractional Brownian motion model. However, recent studies have shown that skin ultrasound signals have multifractal properties. This paper proposes a lognormal multiplicative cascade model to explain this multi-fractality. Experimental results on real images show the validity of the model supported by excellent KS tests. The parameters of the model are proposed as indicators to characterize underlying tissues. Successful application to distinguish melanoma from healthy dermis is presented.