RESUMO
This paper describes the development and initial testing of an automated ultrasound imaging technique to acquire quantitative volumetric breast data; the clinical application being breast cancer diagnosis and management. A novel mechanical scanner has been designed and constructed to constrain the breast tissue without compromising the image, to acquire images of the majority of the breast using a conventional B-mode scanner and to maintain patient comfort. An algorithm to improve upon simple depth-dependent amplification by compensating for tissue-dependent attenuation is applied to the images, making the grey-scale values represent local scattering properties more closely. Registration techniques have been developed to correct for geometric errors arising in the data set because of tissue movement and variations in speed of sound in the tissues. The data sets are reconstructed into volumes and viewed interactively. A pilot study of seven patients was performed and selected results are presented to illustrate lesion features. The automated scan reduces operator-dependence, provides clear information on the 3-D tissue boundaries and provides a full record for monitoring or surgical planning.
