Open AccessResNet based Lung Nodules Detection from Computed Tomography Images
Author(s) -
Mahender G Nakrani*,
Ganesh S Sable,
Ulhas B. Shinde
Publication year - 2020
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.d1540.029420
Subject(s) - artificial intelligence , computer science , lung cancer , nodule (geology) , convolutional neural network , segmentation , stage (stratigraphy) , deep learning , lung , pattern recognition (psychology) , image segmentation , image processing , object detection , radiology , cancer detection , computer vision , cancer , medicine , image (mathematics) , pathology , paleontology , biology
Lung cancer have become one of the major threat to human kind over few years. The survival rate of the patient depends mainly on the stage of cancer when it was detected with early stage detection increases survival rate significantly. Many computer aided detection systems were proposed to assist radiologist in detecting lung nodules efficiently. After the success of deep learning neural network in object classification problem, researchers started adopting it for different tasks in medical image processing and hence in lung nodule detection systems. Hence, a lung nodule detection method using ResNet in CT images is proposed. The proposed method consists of two stages, the pre-processing stage and nodule detection stage. The proposed technique uses morphological operations for segmentation of lungs and convolutional neural network for detection of lung nodules. This method is developed with an aim to provide second opinion to radiologists and reduce their workload. LIDC (Lung Image Database Consortium) dataset which contains 1010 CT patients images of chest regions are taken for experimentation. The model was able to achieve top-5 accuracy of 95.24% on test dataset.