Computer analysis of echographic textures in hashimoto disease of the thyroid

Publisher Website
Google Scholar
Abstract
Ultrasound B-scan images of the thyroid obtained from 10 patients with Hashimoto disease were digitized and processed by a computer method of image analysis that segments complex B-scan images into regions of homogeneous texture. The method was first applied to B-scan images of the normal thyroid and it consistently classified the normal tissue into a unique region. When applied to Hashimoto disease B-scan images, the same method segmented the thyroid into two regions. Detailed analysis of these regions revealed that their gray-level histograms were very different from that of the normal thyroid in eight cases. In two cases the histogram of one of the regions was similar to that of the previous eight cases, whereas the histogram and the tissue of the other region were similar to those of the normal tissue. This paper shows how these results can be interpreted according to the natural history of the Hashimoto disease.