An Improved Faster R-CNN for Pulmonary Embolism Detection From CTPA Images

Computer-aided detection of pulmonary embolism is an important technology method for diagnosing embolism, which can help doctors diagnose quickly and save a lot manpower. However, due to the small area in Computed Tomography Pulmonary Angiography (CTPA) slice images, some previous methods detecting have high number false missed detection. This study proposes based on improved faster region-based convolutional neural network (Faster R-CNN) named More Accurate Faster R-CNN (MA R-CNN). A new feature fusion Multi-scale Fusion Feature Pyramid Network (MF-FPN) proposed by extending adding two bottom-up paths (FPN). It enhances extraction capability entire transmitting low-level accurate location information, makes up original information lost after multiple down-sampling, strengthens use detailed more helpful object. In prediction module, residual block added before fully-connected layer deepen enhance classification accuracy, module (RPM). Compared with R-CNN, MA combines MF-FPN RPM has higher precision solves problems effectively. The average (AP) reached 85.88% CTPA dataset used this article.