ỨNG DỤNG TRÍ TUỆ NHÂN TẠO HỖ TRỢ CHẨN ĐOÁN UNG THƯ PHỔI QUA HÌNH ẢNH CẮT LỚP VI TÍNH LỒNG NGỰC

Abstract

Objective: To evaluate the effectiveness of the artificial intelligence (AI) application developed at Bach Mai Hospital in collaboration with Vietnam National University, Hanoi, in diagnosing lung cancer based on chest computed tomography (CT) images, compared to the assessments of medical experts. Subjects and Methods: A descriptive study involving 200 patients (100 histologically confirmed lung cancer cases, 100 non-cancer cases) at Bach Mai Hospital. Recorded lesion features included location, size, border characteristics, calcification, and cavitation (malignant group); consolidation, ground-glass opacity, and tree-in-bud lesions (benign group). Statistical measures included sensitivity, specificity, accuracy, and receiver operating characteristic (ROC) curve analysis using SPSS 23 (significance level p<0.05). The study was approved by the Biomedical Ethics Committee (733/BM-HĐĐĐ). Results: The AI achieved a sensitivity of 90%, specificity of 91%, and accuracy of 90%. The AI model performed well with benign lesions and solitary pulmonary nodules but had limitations in assessing calcification and cavitation, often confusing tumors with inflammatory lesions, particularly in rare disease forms. Conclusion: AI shows promising potential but still has limitations that require improvement before widespread clinical application.