Aspects of the statistical texture analysis of medical ultrasound images

Although atherosclerotic plaques can be classified into categories characterized by their textures, to visually identify a specific tissue is not an easy task. Statistical texture analysis has therefore been applied. Ten types of statistics are considered: gray level run length; spatial gray level co-occurrence matrices; maximum minimum; Fourier power spectrum; gray level difference; sum and difference histograms; laws texture features; local standard deviation of intensity; organizational approach; and Walsh transform. The effectiveness of these statistics in tissue characterisation is addressed considering polar coordinates calculation as rectilinear coordinates calculation, and taking into account the variation of statistical parameters.<
>