Diagnosis of Acute Appendicitis with Machine Learning-Based Computer Tomography: Diagnostic Reliability and Role in Clinical Management.

Abstract

Purpose: Acute appendicitis (AA) is a common surgical emergency affecting 7-8% of the population. Timely diagnosis and treatment are crucial for preventing serious morbidity and mortality. Diagnosis typically involves physical examination, laboratory tests, ultrasonography, and computed tomography (CT). This study aimed to evaluate the effectiveness of artificial intelligence (AI) in analyzing CT images for the early diagnosis of AA and prevention of complications. Methods: CT images of patients who underwent surgery for AA at the General Surgery Clinic of Kanuni Sultan Suleyman Health Application and Research Center between January 1, 2019, and June 31, 2023, were analyzed. A total of 1200 CT images were evaluated using four different AI models. The model performance was assessed using a confusion matrix. Results: The median age of the patients was 28 years, with a similar sex distribution. No significant differences were observed in terms of age or sex (P = .168 and P = .881, respectively). Among the AI models, MobileNet v2 showed the highest accuracy (0.7908) and precision (0.8203), whereas Inception v3 had the highest F-score (0.7928). In the receiver operating characteristic analysis, MobileNet v2 achieved an area under the curve (AUC) of 0.8767. Conclusion: AI's role in daily life is expanding. In the present study, the highest sensitivity and specificity were 77% and 86%, respectively. Supporting CT imaging with AI systems can enhance the accuracy of AA diagnoses.