Experience of AI deployment at Bach Mai Hospital for the diagnosis of lung cancer

Background: Early detection of lung cancer is difficult. Current tests depend on expert interpretation and may miss subtle lesions.
Approach: At Bach Mai hospital, we developed software that analyzes three data types—chest CT, bronchoscopy, and histopathology. It flags nodules and suspicious regions on CT, recognizes endobronchial lesions and suggests biopsy targets on bronchoscopy, and assists subtype classification on pathology (e.g., adenocarcinoma vs. squamous). An end‑to‑end workflow—from image ingestion to AI analysis and clinician verification—was publicly demonstrated at a national exhibition in Hanoi (28 Aug–5 Sep 2025).
Initial experience: Research and testing indicate faster, more accurate lesion detection and reduced clinician workload. This is the first hospital‑developed AI in Vietnam to cover all three imaging modalities for lung‑cancer diagnosis, with potential to benefit lower‑level facilities.
Conclusion: A hospital‑built, multi‑modal AI pipeline is feasible and promising for accelerating lung‑cancer detection and strengthening equitable care in Vietnam.