Figure 2
From: Automated thorax disease diagnosis using multi-branch residual attention network
Framework of the proposed MBRANet model. The MBRANet architecture consists of a feature extractor, feature fusion, and a multi-branch classifier (MFC). Initially, the input CXR image is fed into the feature extractor, generating four distinct feature scales, labeled as F1, F2, F3, and F4. Subsequently, the feature fusion process is carried out using the FPN technique, resulting in the creation of new features denoted as \(F1^{'}\), \(F2^{'}\), \(F3^{'}\), and \(F4^{'}\). These fused features are then subjected to classification using the MFC approach, resulting in classified feature vectors namely \(y_1\), \(y_2\), \(y_3\), and \(y_4\). Finally, the heat map of the lesion area is obtained, as well as the final prediction \(y_{final}\) using the weighted decision fusion method.
