Tumor location on electroporation therapies by means of multi-electrode structures and machine learning.

Electroporation is a phenomenon produced in the cell membrane when it is exposed to high pulsed electric fields that increases its permeability. Among other application fields, this phenomenon can be exploited in a clinical environment for tumor ablation therapies. In this context to achieve optimum results, it is convenient to focus the treatment on the tumor tissue to minimize side effects. In this work, a pre-treatment tumor location method is developed, with the purpose of being able to precisely target the therapy. This is done by taking different impedance measurements with a multi-output electroporation generator in conjunction with a multi-electrode structure. Data are processed by means of a vector of independent artificial neural networks, trained and tested with simulation data, and validated with phantom gels. This algorithm proved to provide suitable accuracy in spite of the low electrode count compared to the number of electrodes of a standard electrical impedance tomography device.