Electron/pion separation with an Emulsion Cloud Chamber by using a Neural Network
We have studied the performance of a new algorithm for electron/pion separation in an Emulsion Cloud Chamber (ECC) made of lead and nuclear emulsion films. The software for separation consists of two parts: a shower reconstruction algorithm and a Neural Network that assigns to each reconstructed shower the probability to be an electron or a pion. The performance has been studied for the ECC of the OPERA experiment [1]. The $e/π$ separation algorithm has been optimized by using a detailed Monte Carlo simulation of the ECC and tested on real data taken at CERN (pion beams) and at DESY (electron beams). The algorithm allows to achieve a 90% electron identification efficiency with a pion misidentification smaller than 1% for energies higher than 2 GeV.