Evidential Decision Fusion of Deep Neural Networks for Covid Diagnosis

Abstract

In this work, we propose a novel multisource deep learning architecture that employs the evidential Transferable Belief Model (TBM) for combining classifiers for Covid diagnosis. Our architecture was used in the difficult task of distinguishing mild cases of Covid versus severe ones that require urgent medical attention. The available datasets comprised radiographic and clinical data of the patients that we classified separately with a Convolutional Neural Network (CNN) and a decision tree respectively. In our approach, TBM was systematically used to fuse both the results of individual layers in the CNN and to combine the outputs of the CNN with the decision tree. We experimented with both feature and decision fusion approaches. The results outperform the individual classifiers and classical fusion methods.