Lung Nodule Detection using Deep Learning

Principal Investigator

Name: Vismantas Dilys
Degrees: MSc
Institution: Canon Medical Research
Position Title: Scientist
Email: vismantas.dilys@eu.medical.canon

Project Information

Study: NLST
Project ID: NLST-388
Title: Lung Nodule Detection using Deep Learning

Summary

Lung cancer is the leading cause of cancer death worldwide [1]. The National Lung Screening Trial [2] showed a reduction of 20% in lung cancer mortality in high-risk subjects scanned with low-dose Computed Tomography (CT), compared to the control group that received chest radiography. As a consequence of this result, lung cancer screening programs with low-dose CT imaging are being implemented in the US. The implementation of screening would mean a significant increase of reading effort for radiologists.

We propose training of a deep neural networks based CAD system to make lung cancer screening more cost-effective [3-5].

[1] Cancer Facts and Figures 2014 Am. Cancer Soc., 2014 [Online]. Available: http://www.cancer.org/acs/groups/cid/documents/webcontent/003115-pdf.pdf
[2] D. R. Aberle et al., “Reduced lung-cancer mortality with low-dose computed tomographic screening,” N. Eng. J. Med., vol. 365, pp. 395–409, 2011.
[3] White, C.S., Pugatch, R., Koonce, T., Rust, S.W. and Dharaiya, E., 2008. Lung nodule CAD software as a second reader: a multicenter study. Academic radiology, 15(3), pp.326-333.
[4] Sahiner, B., Chan, H.P., Hadjiiski, L.M., Cascade, P.N., Kazerooni, E.A., Chughtai, A.R., Poopat, C., Song, T., Frank, L., Stojanovska, J. and Attili, A., 2009. Effect of CAD on radiologists' detection of lung nodules on thoracic CT scans: analysis of an observer performance study by nodule size. Academic radiology, 16(12), pp.1518-1530.
[5] Yuan, R., Vos, P.M. and Cooperberg, P.L., 2006. Computer-aided detection in screening CT for pulmonary nodules. American Journal of Roentgenology, 186(5), pp.1280-1287.

Aims

• Develop a deep neural networks based CAD system to decet lung nodules.
• Asses system performance in detecting nodules and classifying them as benign or malignant.
• Asses system robustness by using similar data bases for testing such as LIDC-IDRI

Collaborators

Vismantas Dilys, Scientist, Canon Medical Research Europe Ltd.
Erin Beveridge, Clinical Analyst, Canon Medical Research Europe Ltd.
Keith Goatman, Principal Scientist, Canon Medical Research Europe Ltd.