Simulation of dynamic contrast-enhanced ultrasound sequences using example-based texture generation

Abstract

Listen

Translate article

Simulation of dynamic contrast-enhanced ultrasound sequences with known perfusion characteristics and speckle characteristics that are consistent with those observed in experimental data would provide a useful tool for the evaluation of new perfusion quantification and image-processing techniques. A framework is proposed to simulate such perfusion data. It is based on the use of an example-based texture generation method. The generated texture of noise is compared to experimental data in terms of its statistical distribution and spatial correlation. Results show that the example-based method generates data that are closer to the experimental data than those obtained using a conventional parametric simulation method (33 to 80% smaller Hellinger squared distance). This fast and simple method allows simulation of dynamic contrast-enhanced ultrasound data for complex perfusion patterns, and should be useful for the validation of registration, segmentation or perfusion quantification methods.