Application of a Max-Min algorithm on ultrasonic signals for tissue texture analysis

In echographlc imaging, the texture of a lesion is expected to be correlated to the state of the tissue (normal or pathologic). In some cases, the texture is qualitatively described by the ultrasonographer. More quantitative approaches of texture estimation are needed in order to be operator-independent and to allow following-up of lesions. Such approaches have been introduced in the literature but one of the main problems encountered is to obtain tissue parameters independent on experimental conditions: ultrasonic field, Time Gain Compensation (T.G.C), tissue sample attenuation, size of the region of interest. The authors introduce a solution based on the Max-Min algorithm proposed by Mitchell for image analysis. They have shown that the estimated texture parameter is independent on gain, T.G.C, attenuation and size of region of interest. They have also pointed out, through simulation studies, that the parameter depends on one important tissue feature: the number of scatterers with impulse length. This feature is modified in pathological conditions.