文摘
Our in-vivo ultrasonic tissue characterization (UTC) methods are able to successfully analyze diffuse liver disease. The application of these methods to the analysis of metastases in the liver is difficult due to: the wide variety of echographic metastasis appearances, and the small analysis window. A multi-stage metastasis model extended with an automatic core and surrounding halo segmentation method was developed and employed in this study. We performed serial rf-acquisitions of 12 patients with primary adenocarcinoma of the colon. Parenchymal and metastatis UTC features were estimated in semi-automatically delineated regions. The results of over 200 metastases have been analyzed and compared to pseudo metastases delineated in the liver of healthy volunteers. Attenuation could be estimated in only 1 out of 5 metastases, and did not significantly differ from parenchyma. Backscatter intensity is an obvious discriminative feature, but in addition we found an increased backscatter slope with frequency in hypo-echoic metastatic tissue. The texture feature SNR is strongly dependent on the region size, and is not discriminative using a fixed window. We discuss how the results of the backscatter features can be applied to explain several phenomena and show how the results can be used to develop a method for quantitative assessment of the progression of metastases, either or not, due to chemotherapy.