Abstract 607P
Background
An integrated TB and lung cancer screening model was piloted in Vietnam, using artificial intelligence (AI) software to identify potentially malignant nodules on chest X-ray (CXR) when people were initially being screened for TB, in order to indicate further lung cancer screening using a computed tomography (CT) scan.
Methods
From October 2022 to March 2024, CXR images were collected from community-based TB screening events in Ho Chi Minh City (HCMC) and Hai Phong, as well as from individuals undergoing clinical consultation at the Pham Ngoc Thach Hospital in HCMC and the Hai Phong Lung Hospital. CXR images were processed using qXR AI software (Qure.ai, India) to identify those eligible for a CT referral. An on-site radiologist reviewed CT scans, confirmed the presence of malignant nodules and indicated follow-on testing in line with the policies for Vietnam’s social health insurance scheme. Follow-on testing, lung cancer diagnosis and treatment data for people with malignant nodules were exported from each hospitals medical record system.
Results
135,998 people were screened by CXR, resulting in the detection of 1,733 (1.3%) potentially malignant nodules by the AI software. 959 (55.3%) of these individuals were diagnosed with TB or already had a lung cancer diagnosis, leaving 774 (44.7%) eligible for a CT referral. A total of 500 (64.6%) participants got a CT scan, and 316 (84.2%) had radiologist-identified malignant nodules. Follow-up tests were completed for 266 (79.9%) participants, resulting the diagnosis of 133 (50.0%) and treatment of 84 (63.2%) for lung cancer.
Conclusions
AI-assisted CT scan referrals for lung cancer screening were feasible to implement within community- and facility-based TB screening programs. However, community screening suffered from lower yields and more loss in the referral and post-CT scan cascade. Future studies may evaluate the added value of the AI software to detect potentially malignant nodules and early lung cancer detection.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
Friends for International TB Relief (FIT).
Funding
Qure.ai.
Disclosure
A.J. Codlin, T. Dao: Financial Interests, Institutional, Funding, Grant for study: Qure.ai. All other authors have declared no conflicts of interest.