View in EDS

Deep learning-based lung cancer detection using convolutional neural networks.

Saved in:
Bibliographic Details
Title: Deep learning-based lung cancer detection using convolutional neural networks.
Authors: Khattar, Sonam, Aftaab, Mohd, Verma, Tushar, Patial, Deepanshu, Kaur, Bhupinder, San, Hsu Thiri Soe, Kaur, Bhavleen
Source: AIP Conference Proceedings; 2024, Vol. 3072 Issue 1, p1-13, 13p
Subject Terms: LUNG cancer, CONVOLUTIONAL neural networks, DEEP learning, LUNGS, IMAGE databases, DATABASES, RECEIVER operating characteristic curves
Abstract: Lung cancer is one of the most prevalent cancers in the world and is responsible for most cancer-related fatalities. The present methods for identifying lung cancer, such as CT scans and biopsies, have drawbacks, such as being expensive, intrusive, and exposing patients to radiation. Convolutional neural networks (CNN), deep learning, and other machine learning approaches have recently demonstrated promising outcomes in medical image analysis, including lung cancer identification. The openly accessible Lung Image Database Consortium and Image Database Resource Initiative (LIDC-IDRI) used a dataset of 1,010 lung nodules. The nodules in the dataset were both benign and malignant, and they were different sizes and shapes. The photos were preprocessed by being uniformly resized to 32x32 pixels and normalizing the pixel values to be between 0 and 1. The database was separated into a data-training set (80%) and a data-validating set (20%) at random. We built a CNN model with three convolutional layers, each followed by a max-pooling layer and a dropout layer to prevent overfitting. With a batch size of 32 and a learning rate of 0.001, the Adam optimizer was trained to perform model training. The model's sensitivity, specificity, and overall precision on the validation set were 96.6%, 94.7%, and 95.8%, respectively. To assess the accuracy of our CNN model, we compared it to the performance of four radiologists with varying levels of experience in the diagnosis of lung cancer.In terms of sensitivity, our CNN model outscored all four radiologists while matching their specificity levels. AUC of 0.98, showing great diagnostic accuracy, was reached by our CNN model according to a receiver operating characteristic (ROC) analysis we also carried out. Our research shows the promise of CNN models for precise and effective lung cancer nodule detection on CT scans. CNN models can help radiologists identify lung nodules with high accuracy, lowering the possibility of patients being ignored or misdiagnosed. Future research might examine the potential use of Deep networks in clinical decision-making and lung disease detection initiatives. [ABSTRACT FROM AUTHOR]
Copyright of AIP Conference Proceedings is the property of American Institute of Physics and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Complementary Index
Description
Abstract:Lung cancer is one of the most prevalent cancers in the world and is responsible for most cancer-related fatalities. The present methods for identifying lung cancer, such as CT scans and biopsies, have drawbacks, such as being expensive, intrusive, and exposing patients to radiation. Convolutional neural networks (CNN), deep learning, and other machine learning approaches have recently demonstrated promising outcomes in medical image analysis, including lung cancer identification. The openly accessible Lung Image Database Consortium and Image Database Resource Initiative (LIDC-IDRI) used a dataset of 1,010 lung nodules. The nodules in the dataset were both benign and malignant, and they were different sizes and shapes. The photos were preprocessed by being uniformly resized to 32x32 pixels and normalizing the pixel values to be between 0 and 1. The database was separated into a data-training set (80%) and a data-validating set (20%) at random. We built a CNN model with three convolutional layers, each followed by a max-pooling layer and a dropout layer to prevent overfitting. With a batch size of 32 and a learning rate of 0.001, the Adam optimizer was trained to perform model training. The model's sensitivity, specificity, and overall precision on the validation set were 96.6%, 94.7%, and 95.8%, respectively. To assess the accuracy of our CNN model, we compared it to the performance of four radiologists with varying levels of experience in the diagnosis of lung cancer.In terms of sensitivity, our CNN model outscored all four radiologists while matching their specificity levels. AUC of 0.98, showing great diagnostic accuracy, was reached by our CNN model according to a receiver operating characteristic (ROC) analysis we also carried out. Our research shows the promise of CNN models for precise and effective lung cancer nodule detection on CT scans. CNN models can help radiologists identify lung nodules with high accuracy, lowering the possibility of patients being ignored or misdiagnosed. Future research might examine the potential use of Deep networks in clinical decision-making and lung disease detection initiatives. [ABSTRACT FROM AUTHOR]
ISSN:0094243X
DOI:10.1063/5.0198679