Convolutional neural networks for lung cancer screening in computed tomography (CT) scans

Diagnosis and cure of cancer has been one of the biggest challenges faced by mankind in the last few decades. Early detection of cancer would facilitate in saving millions of lives across the globe every year. This paper presents an approach which uses a Convolutional Neural Network (CNNs) to classify tumours seen in lung cancer screening computed tomography scans as malignant or benign. CNNs have special properties such as spatial invariance, and allow for multiple feature extraction. When such layers are cascaded, leading to Deep CNNs, it has been shown widely that the accuracy of prediction increases dramatically. In this work, we have designed a CNN suitable for the analysis of CT scans with tumours, using domain knowledge from both medicine and neural networks. The results show that the accuracy of classification for our network performs better than both the traidtional neural networks, and also existing CNNs built for image classification purposes.