A COMPARATIVE ANALYSIS OF ARTIFICIAL INTELLIGENCE-BASED IMAGE INTERPRETATION VERSUS CONVENTIONAL RADIOLOGIST ASSESSMENT IN THE DIAGNOSIS OF PULMONARY NODULES ON CHEST CT SCANS

Abstract

Artificial Intelligence (AI) has taken over diagnostic radiology as one of the ways through which pulmonary nodules can be detected on chest computed tomography (CT) scans which play a vital role in early lung cancer detection. The study puts forth an elaborate comparative evaluation of AI-driven image interpretation in comparison with standard radiologist testing to assess pulmonary nodules in a population group of 480 patients. When using a deep learning model trained on voluminous annotated productions, the AI system showed better sensitivity than thoracic radiologists who completed their (post graduate fellowship)  (94.2%  v 91.7%), but processed cases faster 1.7 minutes per case compared to the mean 6.4-minute cases. A different matter is that radiologists were more specific (93.1%) than AI (89.5%), which demonstrates strengths. Analysis of errors showed that AI had a higher rate of false positive outcomes, especially when scanning a vascular artifact but EXPANSE radiologists had more difficulties with finding small subsolid nodules. The consistency of the performance of AI was significantly higher than those of the radiologists (AI 0.88 vs. radiologists 0.720.79). AI was also found to have much higher sensitivity in finding small and peripherally located nodules with stratified analysis by the size of nodule and region of the lung. This evidence shows the promise of AI to increase diagnostic accuracy and make the workflow in thoracic imaging more efficient, particularly when applied in hybrid mode with human analysis. This research supports the explanations that promote AI not as a substitute but as an imperative add-on in contemporary radiology practice.

Keywords: Artificial intelligence, pulmonary nodules, chest CT, radiology, deep learning, diagnostic accuracy, radiologist comparison, medical imaging, computer-aided diagnosis, lung cancer screening