Model-based probability of detection of pathologies in soft tissue

This contribution is aimed at presenting a computational method for assessment and prediction of mechanical property changes in soft tissue, in order to diagnose certain pathologies. This diagnostic technology is based on two components: a custom designed piezoelectric transducer that transmits and receives torsion ultrasound, and a model-based inverse problem that simulates using finite elements the ultrasound propagation through the transducer-tissue system and reconstructs the unknown pathology-related mechanical properties using inverse problem-based search algorithms. As a contribution, this work proposes a semi-analytical estimator of the probability of detection (POD) in the sense of the probability that the pathology effect on the measurement overcomes the measurement noise, based on the finite element model of the ultrasound propagation through the transducer-tissue system. The estimator of the POD can be used as a optimality criterion to optimize the design of the transducer. This work aims to (i) evaluate the optimal piezoelectric transducer design of the model-based on Probability of detection. A second goal is (ii) a simulated experiment based on the three dimensional model of wave propagation generated by the proposed piezoelectric transducer design. Finally, (iii) a parametric study is carried out to extract practical parameters for final soft tissue applications. Acknowledgements: Spanish Ministerio de Economia for project DPI201017065, Junta de Andalucia for projects P11-CTS-8089 and GGI3000IDIB, and the European Union for program ’Programa Operativo FEDER de Andalucía 2007-2013’.