Fast recurrent neural network approach for automatic detection of lung nodules

The cancer death in concern to lung cancer is increased compared to other cancers worldwide. The presence of nodules in the lung indicates the chances of getting lung cancer in the future. In this paper, an accurate Computer-Aided Detection (CAD) system for lung nodules detection using Computer Tomo...

Full description

Saved in:
Bibliographic Details
Published in:AIP conference proceedings 2022-10, Vol.2519 (1)
Main Authors: Pattnaik, Balachandra, Suseela, B., Babu, T. R. Ganesh, Anitha, S.
Format: Article
Language:English
Subjects:
Computed tomography
Image segmentation
Lung cancer
Medical imaging
Modules
Nodules
Preprocessing
Recurrent neural networks
Vessels
Wiener filtering
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The cancer death in concern to lung cancer is increased compared to other cancers worldwide. The presence of nodules in the lung indicates the chances of getting lung cancer in the future. In this paper, an accurate Computer-Aided Detection (CAD) system for lung nodules detection using Computer Tomography (CT) images is proposed. It consists of four important modules such as preprocessing; Two-Successive Segmentation Process (TS2P), Rule-Based Refinement Pass (RBRP), and Detection Module (D.M.DM). The lung C.T. CT image is de-noised using the Wiener filter in the preprocessing module. In the TS2P module, the right and left lungs are segmented at first, and in the next stage, the nodules and vessels are segmented. Then, the RBRP module is designed to remove the vessels with the help of geometrical features. Finally, the nodules are detected using a deep learning approach in the D.MDM. The proposed method is validated on 888 lung C.T.CT images, and a mean average precision of 96.75% and sensitivity of 97% with 2 false positives per image were obtained.
ISSN:0094-243X
1551-7616
DOI:10.1063/5.0111863