Artificial Intelligence-Assisted Diagnostic Accuracy in Early-Stage Pancreatic Cancer Detection: A Comparative Analysis

Background: Pancreatic ductal adenocarcinoma (PDAC) remains one of the most lethal malignancies, with five-year survival rates below 10% largely due to late diagnosis. Artificial intelligence (AI) and machine learning algorithms have shown promise in enhancing early detection capabilities across various imaging modalities.

Objective: To evaluate the diagnostic accuracy of deep learning-based AI algorithms in detecting early-stage pancreatic cancer from computed tomography (CT) and magnetic resonance imaging (MRI) scans compared to conventional radiological assessment.

Methods: This multicenter retrospective study analyzed 2,847 imaging studies (1,623 CT scans and 1,224 MRI scans) from patients across three continents between January 2019 and December 2023. A convolutional neural network (CNN) was trained on 70% of the dataset and validated on the remaining 30%. Performance metrics including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under the receiver operating characteristic curve (AUROC) were calculated and compared with interpretations from experienced radiologists.

Results: The AI algorithm demonstrated a sensitivity of 91.3% (95% CI: 88.7-93.5%) and specificity of 94.8% (95% CI: 93.1-96.2%) for detecting early-stage PDAC, compared to 84.2% (95% CI: 81.1-87.0%) and 89.6% (95% CI: 87.5-91.4%) respectively for radiologist interpretation (p<0.001). The AUROC for AI was 0.961 versus 0.893 for conventional reading. AI algorithms identified 127 additional true positive cases that were initially missed on routine radiological review, representing a 15.1% improvement in early detection rates.

Conclusion: AI-assisted diagnostic tools significantly enhance the detection of early-stage pancreatic cancer, offering potential for improved patient outcomes through earlier therapeutic intervention. Integration of these algorithms into clinical workflows warrants further prospective evaluation.

Keywords: Pancreatic cancer, artificial intelligence, machine learning, diagnostic imaging, early detection, convolutional neural network