Lung cancer is the result of the abnormal division and growth of airway epithelial cells, mainly in the epithelium of the bronchi. The process of dividing cells quickly goes out of control and creates tumors. Lung cancer interferes with lung function, spreading cancer cells in the adjoining, lymphatic or bloodstream to nearby lymph nodes or to other parts of the body.
According to recent statistics by the International Agency for Research on Cancer (IARC) and the GLOBOCAN database, lung cancer is the leading cause of death in both men and women around the world. In Vietnam, in 2018, lung cancer ranks second in both new cases (23,667 cases, accounting for 14.4% of all new cases) and deaths (20,710 cases, accounting for 18.0% of deaths). Lung cancer is divided into two main types, small cell lung cancer, accounting for 15% and non-small cell lung cancer, accounting for 85%.
Today with the strong development of artificial intelligence, the health sector has been promoting the application of information technology in medical examination and treatment activities, and software systems to support doctors in diagnosing diseases is an inevitable trend. In Vietnam, the software system to support diagnosis not only helps to reduce the workload for physicians in central hospitals, but also assists doctors at the provincial level and the doctors in remote areas to diagnose accurately and effectively.
Stemming from such practical needs, VAST assigned a project "Building a medical image recognition system to support diagnosis of lung cancer based on machine learning and a high-performance computing platform" to the University of Science and Technology of Hanoi (USTH) to coordinate with K Hospital, implemented from January 2018 to December 2019. The project belongs to the application of information technology in health, chaired by Dr. Tran Giang Son. Authors focused on non-small cell lung cancer.
Result image of nodular recognition model, lung tumor on computerized tomography image
Authors have successfully built a model to detect and identify the location, size of nodules, lung tumors on computerized tomography, and successfully built a model of nodule classification, lung tumor on scan images on computerized tomography that is benign or malignant. In addition, the authors have also developed computerized tomography imaging data sets for lung cancer in Vietnam.
The team of scientists who implemented the project from the University of Science and Technology of Hanoi said that the development of a toolkit to assist doctors in analyzing and labeling computerized tomography image data on lung cancer built a system of image recognition software that is a first step in assisting doctors in detecting and identifying nodules and lung tumors on computerized tomography images, helping reduce the workload for doctors at central hospitals.
Other results of the project include training for 01 post-graduate student at University of Science and Technology of Hanoi, successful training for 01 graduate student at University of Science and Technology of Hanoi (successfully defended master thesis in October 2019); publishing 02 articles on journals in the list of SCIE.
The project has been assessed by VAST level as excellent on February 14th, 2020./.
Translated by Tuyet Nhung
Link to Vietnamese version