Neuro-chip design for developing wearable medical diagnostic E-sniffer

The diagnosis of infection caused by E.Coli seems to be one of the primary pre-requisites of successful medical treatment and as such is a high priority in clinical science for the diseases like Urinary Tract Infections and Sepsis. Researchers have proved that diagnosis of microbial infections can be improved by E-sniffer technology which consists of Artificial Neural Network as pattern recognizer. As on date the neural network implementations are in software, hence such technology suffers from portability, manipulation and high cost. With the appearance of large, dense, highly parallel Field Programmable Gate Array circuits, it has now become possible to realize large-scale neural networks in hardware, with the flexibility and low cost of software implementations. This paper aims at developing the Artificial Neural Network as neuro chip so that portable medical diagnostic E-sniffer for E.Coli identification can be prototyped. The Multilayer perceptron with 12 input neurons, one hidden layer with 8 neurons and 3 output neurons are made hardware realizable in Altera cyclone II DE2 board with NIOSII processor and the EP2C35F672C6 as target device. The functional simulation is done using Modelsim 8.5C and cross compared with the software solutions obtained using Nerosolution4.2. After implementation the circuit is verified for its functionality and timing. The performance in terms of sensitivity and specificity for this implementation are highly satisfactory.