Journal of Ultrasound in Medicine, Vol 20, Issue 5 465-471, Copyright © 2001 by American Institute of Ultrasound in Medicine

JOURNAL ARTICLE

Ultrasonographic texture analysis of parenchymatous organs by the four-neighborhood-pixels algorithm: clinical experiment

W. Yao, B. Zhao, Y. Zhao, W. Wang and G. Qian
Ultrasonic Examination Department, Changhai Hospital, Shanghai, People's Republic of China.

OBJECTIVE: The parenchyma of organs such as liver, thyroid, and mammary gland during climacterium have common ultrasonographic textural features, which together form what we call small-dot-structure texture. To study this texture we designed the 4-neighborhood-pixels algorithm, an ultrasonographic texture analysis algorithm. The objective of this study was to confirm whether the 4-neighborhood-pixels algorithm can reflect the features of small-dot-structure texture. METHODS: A changed small-dot-structure texture and 3 other textures were compared with the normal small-dot-structure texture in 4 groups, and a histogram algorithm was used for contrast with the 4-neighborhood-pixels algorithm. RESULTS: The 4-neighborhood-pixels algorithm could reflect all the textural differences, but the histogram algorithm could reflect only some of them. CONCLUSIONS: The 4-neighborhood-pixels algorithm is a good algorithm for analyzing ultrasonographic small-dot-structure texture. Not only can it reflect changes in the small-dot-structure texture, but it can also differentiate between small-dot-structure and non-small-dot-structure textures.