Lung Nodule Classification Based on Deep Convolutional Neural Networks

Author(s)

Mendoza Bobadilla, Julio Cesar ; Pedrini, Helio ; BeltranCastanon, C ; Nystrom, I ; Famili, F

• Document type: Journal article
• Source: PROGRESS IN PATTERN RECOGNITION, IMAGE ANALYSIS, COMPUTER VISION, AND APPLICATIONS, CIARP 2016; v. 10125, p. 8-pg., 2017-01-01.

Abstract

Lung nodule classification is one of the main topics on computer-aided diagnosis (CAD) systems for detecting nodules. Although convolutional neural networks (CNN) have been demonstrated to perform well on many tasks, there are few explorations of their use for classifying lung nodules. In this work, we present a method for classifying lung nodules based on CNNs. Training is performed by balancing the mini-batches on each stochastic gradient descent (SGD) iteration to address the lack of nodule samples compared to background samples. We show that our method outperforms a base feature-engineering method using the same techniques for other stages of lung nodule detection, and show that CNNs obtain competitive results when compared to state-of-the-art methods evaluated on Japanese Society of Radiological Technology (JSRT) dataset [13]. (AU)

• FAPESP's process: 11/22749-8 - Challenges in exploratory visualization of multidimensional data: paradigms, scalability and applications
• Grantee: Luis Gustavo Nonato
• Support Opportunities: Research Projects - Thematic Grants