Detection of pulmonary nodules is a dangerous kind of lung cancer that is responsible for majority of deaths every year. Early diagnosis and proper treatment of Pulmonary Nodules significantly improves the patient’s survival rate. In this study, we propose a multi-view convolutional network for pulmonary nodule detection. The main objective of our work is to establish a method that can automatically pre-process, localize and then segment the pulmonary nodules precisely and improve its accuracy. In our proposed method single shot multi-box detector (SSD) precisely localizes the nodules area in the form of bounding boxes and eliminates some clinical artifacts. The proposed approach was evaluated on LUNA 2016 dataset to show the robustness of our work which achieved a sensitivity and precision of 97.47 and 0.97 respectively. The results of the segmented image are also compared with the state-of-the-art methods to demonstrate the performance superiority of the proposed approach.
Muhammad Abdullah, Aun Irtaza, Aun Irtaza, Nudrat Nida. (2019) Lungs Malignancy evaluation of the pulmonary nodules using deep learning, , Proc. of the PAS: A; 56, issue 4.
