Microwave tomography with phaseless data on the calcaneus by means of artificial neural networks

LOTTO, F.P.; IRASTORZA, R.M.; VERICAT, F.; FAJARDO, J.E.; CARLEVARO, C.M.

MEDICAL AND BIOLOGICAL ENGINEERING AND COMPUTING

SPRINGER HEIDELBERG

Lugar: HEIDELBERG; Año: 2019

0140-0118

The aim of this study is to use a Multilayer Perceptron (MLP) ArtificialNeural Network (ANN) for phaseless imaging the human heel (modeled as a bilayerdielectric media: bone and surrounding tissue) and the calcaneus cross-sectionsize and location using a two dimensional (2D) microwave tomographic array.Computer simulations were performed over 2D dielectric maps inspired byComputed Tomography (CT) images of human heels for training and testing theMLP. A morphometric analysis was performed to account for the scatterer shapeinfluence on the results. A robustness analysis was also conducted in order tostudy the MLP performance in noisy conditions. The standard deviations of therelative percentage errors on estimating the dielectric properties of thecalcaneus bone were relatively high. Regarding the calcaneus surroundingtissue, the dielectric parameters estimations are better, with relativepercentage error standard deviations up to ≈ 15 %. The location andsize of the calcaneus are always properly estimated with absolute errorstandard deviations up to ≈ 3 mm