Deep learning model for discriminating early lung cancer from TB nodule and risk assessment management for lung nodule
Principal Investigator
Name
SANDRA VINAY KUMAR
Degrees
Ph.D (III Year)
Institution
National Institute of Technology ,Karnataka
Position Title
Ph.D Scholar
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-1013
Initial CDAS Request Approval
Feb 8, 2023
Title
Deep learning model for discriminating early lung cancer from TB nodule and risk assessment management for lung nodule
Summary
Lung cancer is the most common cause of cancer-related death worldwide. Most lung cancers (85–90%) are classified as non-small cell lung cancer (NSCLC), which is highly correlated with smoking and has a survival rate dramatically affected by the stage at detection.
In contrast to developed nations where the incidence of smoking is falling, cigarette smoking is on the rise in developing nations, many of which have a high rate of endemic tuberculosis.
In low and middle-income countries with both high rates of smoking and endemic tuberculosis, identification of early lung cancer can be significantly confounded by the presence of lung nodules due to latent TB (LTB). Unfortunately, these two entities cannot be readily distinguished, even by trained radiologists. This diagnostic equipoise leads to significant delays in cancer diagnosis, a disease for which timely intervention is paramount, with concomitant increases in lung cancer mortality.
We are going to develop a deep learning model which will detect and classify the presence of pulmonary nodules in CT scans and give a risk scoring base on the likelihood of malignant or benign. as well as distinguishing between TB and lung cancer.
Need to access the NLST data to further our ongoing research and development work in distinguishing between TB and lung cancer.
In contrast to developed nations where the incidence of smoking is falling, cigarette smoking is on the rise in developing nations, many of which have a high rate of endemic tuberculosis.
In low and middle-income countries with both high rates of smoking and endemic tuberculosis, identification of early lung cancer can be significantly confounded by the presence of lung nodules due to latent TB (LTB). Unfortunately, these two entities cannot be readily distinguished, even by trained radiologists. This diagnostic equipoise leads to significant delays in cancer diagnosis, a disease for which timely intervention is paramount, with concomitant increases in lung cancer mortality.
We are going to develop a deep learning model which will detect and classify the presence of pulmonary nodules in CT scans and give a risk scoring base on the likelihood of malignant or benign. as well as distinguishing between TB and lung cancer.
Need to access the NLST data to further our ongoing research and development work in distinguishing between TB and lung cancer.
Aims
Development and testing of our deep learning model for pulmonary nodule detection and give a risk scoring and discriminating early lung cancer from TB nodule.
Collaborators
Munyanaik Kethavath, NITK,INDIA