DIAGNOSIS AND PROGNOSIS OF LUNG CANCER BY AI-ASSISTED COMPUTED TOMOGRAPHY

Abstract

Lung cancer is one of the most common and lethal cancers in the world, and is a significant challenge when diagnosed late. Computed tomography (CT) plays an important role in early detection, but its routine use requires caution due to its cost, radiation exposure, and divergence from international guidelines. Artificial intelligence (AI) has emerged as a promising tool for improving CT, but its application must follow clinical and scientific criteria for better use and greater safety. This article is a literature review conducted in the PubMed database between March and April 2025, using the descriptors “CT”, “AI”, and “Lung cancer diagnosis”. After applying the inclusion and exclusion criteria, 4 articles were selected to compose the final analysis, being complemented by primary references, in order to investigate the use of artificial intelligence in the diagnosis of lung cancer. AI has demonstrated great potential in early detection, tumor characterization, and staging, overcoming limitations of human analysis and increasing diagnostic accuracy. Furthermore, it has enabled the integration of clinical and genetic data, improving the prediction of risk and clinical outcomes. AI-based models already show high accuracy in differentiating between benign and malignant nodules, including the generation of synthetic PET images similar to real ones. The technology shows promise, but still requires additional scientific validation and improvement, with regard to the unique attributes of human capacity. Therefore, this review discusses how artificial intelligence, integrated with computed tomography, represents a promising advance in the early diagnosis of lung cancer, offering greater sensitivity, specificity and reproducibility, contributing not only to detection, but also to characterization, staging, disease monitoring, and optimization of patient prognosis.