An artificial neural network hardware for bladder cancer

Abstract

An Artificial Neural Network was implemented on a FPGA to analyze electrical impedance spectroscopy (EIS) data taken from a patient bladder. Using this system, malignant areas from non-malignant areas in the urinary bladder of a patient can be separated very rapidly. Two variants of backpropagation algorithm were used to train a multilayer perceptron (MLP) neural network. By adjusting the ANN's parameters, the maximum error and error percentage of test phase in 300 epochs were reduced to 0.33297 and 0.39035%, respectively. The ANN architecture has been implemented on a Virtex-4 LX25 FPGA from Xilinx. The number of occupied slices on the FPGA is 10136 and the design covers 94% of the chip. آ© EuroJournals Publishing, Inc. 2009.