Artificial Neural Network Model for Mass Characterization in Breast Palpation

In this paper an artificial neural network model for identifying inclusion properties from palpation experimental data is demonstrated. The forward model of breast palpation is based on the combination of biomechanics model and experimental data. The loading-displacement curve displays the combination of two components: Gaussian and exponential components. The standard deviations and amplitudes of its Gaussian component are related to inclusion properties. Exponential component is well explained by soft tissue indentation theory. The inverse problem of soft tissue palpation is solved using an artificial neural network (ANN) model. To obtain a data basis for the training and validation of the artificial neural network, experiments were carried out for different sets of inclusion parameters. The results show that the ANN model has the capability to predict the inclusion properties when the indentation depth is close to the underlying depth of the inclusion.