Factors Affecting the Statistical Analysis of Ultrasonic Backscattering Signals and Imaging
Statistical distribution of ultrasonic backscattering signals has been demonstrated capable of characterizing variations of density and arrangement of scatterers in biological tissues. The statistical analysis of ultrasound signals has also found with less dependency on the attenuation effect. Yet, as the employed ultrasound frequency and pulse duration were increased, several factors could further affect the precise estimation of the statistical parameters. To further investigate the addressed issues, experiments were arranged and performed from tissue-mimicking phantoms and porcine livers. Various duty cycles, including 1, 3, 5, and 10%, of tone bursts at 1 KHz pulse repetition frequency corresponding to ultrasound frequencies of 3.5, 7.5, and 10 MHz were adjusted for driving the transducers. The tissue-mimicking phantoms were fabricated, which consisted of gelatin and glass beads of 16 and 64 scatterers/ mm3 and those of porcine livers with either healthy or pathological fibrosis were obtained from local slaughter house. Various thickness of Silicone plates with the attenuation coefficient of 1.62 dB/mm..MHz were placed on the surface of objects to be measured. Nakagami statistical model, including shape parameter (Nakagami-m) and parametric imaging, was implemented to assess variations of the probability density function (PDF) estimated from the acquired ultrasonic backscattering signals. Results of phantoms indicated that the attenuation could significantly vary the shape of PDF of backscattered envelopes. Especially, large attenuation effect was found corresponding to those broader incident ultrasound bandwidths excited by the monocycle signal; whereas the effect is substantially reduced as the tone bursts were more than 3 cycles. Results of porcine livers indicated that the Nakagami-m increased with the increasing ultrasound frequencies and bandwidth, and that those associated PDFs were nearly pre- Rayleigh distributed. Both phantoms and porcine livers results consistently demonstrated that the use of 3 cycles tone bursts for exciting transducers may achieve the most appropriate performance to accommodate a tradeoff between attenuation effect and image resolution. Current study also verified that the operational modes of incident ultrasound need to be properly assured before that the statistical model may be further applied to clinical applications.
KeywordsUltrasound backscattering Statistical model Probability distribution function parametric imaging
Unable to display preview. Download preview PDF.