1Department of Electrical Engineering, Shonan Institute of Technology, 2Institute of Information Sciences and Electronics, Usiversity of Tsukuba, 3Department of Clinical Pathology, Jichi Medical School
This paper describes the fractal analysis of a B-mode image for tissue characterization. We consider that
if a self similarity can be assumed in structures of the tissues, the ultrasonic images will be affected by the
structures dependent on the pathological states. We calculated the Hurst coefficients of B-mode images in
cases of normal and fatty liver, and cirrhosis of the liver. Using 10 MHz ultrasonic waves for rat's liver in vitro
we observed that the relationship of the Hurst coefficient was fatty<normal<cirrhosis. Using 3.5 and 5 MHz
ultrasonic waves for human liver we observed a similar relationship, although the difference between them was
very small, since image degradation occurred due to the use of a lower frequency and subcutaneous fat. Thus,
in thyroid cases with normal, Hashimoto, and hyperthyroiditism, we used 10 MHz ultrasonic waves to
calculate the Hurst coefficients, the relationship between which was normal<Hashimoto<hyperthyroiditism.
Consequently, we found that the Hurst coefficient indicating a degree of self similarity is one of the parameters
for characterizing tissue structures. For the relationship between Hurst coefficients of the B-mode images and
tissues structures, further study is required.