Experimenting with Convolutional Neural Network architectures for the automatic characterization of Solitary Pulmonary Nodules' malignancy rating

Lung cancer is the most common cause of cancer-related death worldwide. Early and automatic diagnosis of Solitary Pulmonary Nodules (SPN) in Computed Tomography (CT) chest scans can provide early treatment as well as doctor liberation from time-consuming procedures. Deep learning has been proved as a popular and influential method in many medical imaging diagnosis areas. In this study, we consider the problem of diagnostic classification between benign and malignant lung nodules in CT images, derived from a PET/CT scanner. More specifically, we intend to develop experimental Convolutional Neural Network (CNN) architectures and conduct experiments, by tuning their parameters, to investigate their behavior, and define the optimal setup for the accurate classification. For the experiments, we utilize PET/CT images obtained from the Laboratory of Nuclear Medicine of University of Patras, and the public database called Lung Image Database Consortium Image Collection (LIDC-IDRI).  Furthermore, we apply simple data augmentation to generate new instances and inspect the performance of the developed networks. Classification accuracy of 91% and 93% on PET/CT and on a selection of nodule images from the LIDC-IDRI datasets is achieved accordingly. The results demonstrate that CNNs are a trustworthy method for nodule classification. Also, the experiment confirms that data augmentation enhances the robustness of the CNNs.