Lung nodule classification by jointly using visual descriptors and deep features

Date

2017

Authors

Xie, Y.
Zhang, J.
Liu, S.
Cai, W.
Xia, Y.

Editors

Muller, H.
Kelm, B.M.
Arbel, T.
Cai, W.
Cardoso, M.J.
Langs, G.
Menze, B.
Metaxas, D.
Montillo, A.
Wells, W.M.
Zhang, S.
Chung, A.C.S.
Jenkinson, M.
Ribbens, A.

Advisors

Journal Title

Journal ISSN

Volume Title

Type:

Conference paper

Citation

Lecture Notes in Artificial Intelligence, 2017 / Muller, H., Kelm, B.M., Arbel, T., Cai, W., Cardoso, M.J., Langs, G., Menze, B., Metaxas, D., Montillo, A., Wells, W.M., Zhang, S., Chung, A.C.S., Jenkinson, M., Ribbens, A. (ed./s), vol.10081, pp.116-125

Statement of Responsibility

Yutong Xie, Jianpeng Zhang, Sidong Liu, Weidong Cai, and Yong Xia

Conference Name

Medical Image Computing and Computer-Assisted Intervention (MICCAI) (21 Oct 2016 - 21 Oct 2016 : Athens, Greece)

Abstract

Classifying benign and malignant lung nodules using the thoracic computed tomography (CT) screening is the primary method for early diagnosis of lung cancer. Despite of their widely recognized success in image classification, deep learning techniques may not achieve satisfying accuracy on this problem, due to the limited training samples resulted from the all-consuming nature of medical image acquisition and annotation. In this paper, we jointly use the texture and shape descriptors, which characterize the heterogeneity of nodules, and the features learned by a deep convolutional neural network, and thus proposed a combined-feature based classification (CFBC) algorithm to differentiate lung nodules. We have evaluated this algorithm against four state-of-the-art nodule classification approaches on the benchmark LIDC-IDRI dataset. Our results suggest that the proposed CFBC algorithm can distinguish malignant lung nodules from benign ones more accurately than other four methods.

School/Discipline

Dissertation Note

Provenance

Description

Access Status

Rights

© Springer International Publishing AG 2017

License

Grant ID

Call number

Persistent link to this record