Background: Artificial intelligence (AI) based automatic segmentation has demonstrated substantial use in measuring features of abdominal aortic aneurysm (AAA) on computed tomography angiography (CTA) scans. Automatic segmentation models have the potential to aid clinicians in assessment and management of AAA patients and save time taken for manual measurement. Objective: This review assessed the performance and accuracy of AI automatic segmentation models in measuring features of AAA. Methods: Electronic databases were searched systematically for studies assessing measurement of AAA features on CTA images using automatic segmentation methods as well as manual segmentation methods. The primary outcome measure of AI model performance was DICE similarity coefficient (DSC), representing the proportional overlap between the prediction segmentation and the standard reference. Secondary outcomes were sensitivity and specificity of AAA diagnosis. Results: A total of 23 studies involving 4015 CTA scans from 1802 patients were included. Meta analysis showed overall pooled mean (95% CI) DSCs for segmentation of AAA, lumen and ILT were 0.90 (0.86-0.94; I2=98.4%, 0.93 (0.92-0.93; I2=98%) and 0.85 (0.82-0.88; I2=96.5%) respectively. Meta-analysis of sensitivity of diagnosis of AAA produced a pooled diagnostic odds ratio (DOR) of 176.1. Conclusion: AI-based models demonstrate diagnostic accuracy for AAA comparable to that of experienced vascular surgeons. Moreover, these models provide enhanced reproducibility across all patients and longitudinal imaging studies due to the consistent application of standardised algorithms.