Quantitative tomography based on ultrasound: aplications in mamography

Abstract

The motivation for this project comes from the literature that has shown that acoustic properties of tissues obtained by ultrasound have high sensitivity for detection of tumors and high specificity in distinguishing between fluid and solid cysts. The use of acoustic radiation, in particular, is interesting, as it is not ionizing radiation; the test can be repeated several times; the equipment presents relatively low cost; and depending on the frequency used, has good resolution and good penetration (tens of centimeters). However, the acoustic properties cannot be estimated by conventional ultrasound (B-mode), applied in clinics and hospitals because it is essentially qualitative. Quantitative tomography is necessary based on distributed sensors around objects of interest.The current tomographic reconstruction methods using distributed transducers are applicable to low and moderate contrast of the acoustic properties. However, there are human tissues that differ substantially from other tissues, both in speed of propagation, material density, as well as in attenuation. Under these conditions, the accuracy of reconstruction is compromised. The hypothesis that we intend to investigate is whether information obtained by reflection of the ultrasound waves can improve quantitative reconstruction based on scattered waves, particularly in dense breasts. The main objective of this project is to explore both the reflected and scattered waves to allow tomographic reconstruction of tissue mechanical properties even in the presence of structures with high contrast. Once identified and delimited, these organs can be initialized with mechanical properties reported in the literature and used as priors in quantitative methods of reconstruction.The evaluation will be based on numerical simulations with increasing level of noise, and on physical phantoms with homogenous and heterogeneous material. The metric will be based on the normalized mean square error and the maximum error between the estimated parameters and predefined parameters. (AU)