Quaternion Neural Network with Temporal Feedback Calculation: Application to Cardiac Vector Velocity during Myocardial Infarction

CRUCES P.D.; CORREA, R.; LACIAR E.; ARINI P.D.

Congreso; XXI Congreso Argentino de Bioingeniería, X Jornadas de Ingeniería Clínica; 2017

Quaternion neural networks have been shown tobe useful in image and signal processing applications. Herein, we propose anovel architecture of a neural unit model characterized by its ability ofencoding 3-dimensional past information and that facilitates the learning ofvelocity patterns. We evaluate the implementation of the network in a study ofthe cardiac vector velocity and its usefulness in early detection of patientswith anterior myocardial infarction. Experimental results show an improvementof the performance in terms of convergence speed and precision when comparingwith traditional methods. Furthermore, the network shows successful results inmeasuring the velocity reduction that is usually observed in vectorcardiogramsignals in the presence of myocardial damage. Through a linear discriminantanalysis, a pair of 100% / 98% of sensitivity / specificity is met with onlytwo velocity parameters. We conclude that this method is a very promisingdevelopment for future computational tools devoted to early diagnosis of heartdiseases.