Improvement of the Precision of Ultrasonic Microscope for Biological Tissue Using the Automatic Extraction of the Reference Signals

This paper deals with the precision of the ultrasonic microscopy for biological tissue characterization. The estimation error of sound speed influenced by the selection of reference points for obtaining the reference signal is described. In well known scanning type microscope, a focused ultrasonic signal is irradiated to the glass substrate on which a thin sliced tissue is attached. The reflected signal from the front surface of the tissue is compared with the signal from the glass substrate on which no tissue is attached. As the scanning plane and glass substrate are not completely in parallel, it is necessary to locate the glass surface at each measuring point. For this purpose, several reference points on which no tissue is attached are selected in the view field. The equation of the plane representing the glass surface is consequently obtained using the reflection at these reference points. However, the glass surface has an apparent surface roughness. This brings an estimation error of the equation of the plane. This error can be reduced by increasing the number of reference points. In order to extract a sufficient number of reference points automatically, an algorithm using the reflected wave was proposed to discriminate the points where the glass surface is exposed. Although the estimation error of sound speed with three reference points was as large as 31 m/s (as the double of standard deviation), it was reduced into as small as 16 m/s, when 1000 reference points had been automatically extracted.