Search

Objective To explore the clinical value of artificial intelligence (AI) in the diagnosis of intracranial aneurysms by head and neck CT angiography (CTA). Methods A total of 117 patients with clinically suspected intracranial aneurysms admitted in our hospital from August 2020 to January 2021 were enrolled, and their head and neck CTA and digital subtraction angiography (DSA) data were collected and retrospectively analyzed. Taking DSA results as the gold standard, the sensitivity and accuracy of AI in the diagnosis of intracranial aneurysms were calculated. In addition, the results of AI automatic measurement were recorded, and the consistency test was conducted with the results of DSA manual measurement. Results Among the 117 suspected patients, there were 105 cases diagnosed with intracranial aneurysms by DSA, with a total of 136 aneurysms. ①Diagnostic efficiency: The sensitivity and accuracy of AI for aneurysm diagnosis were 87.6% and 79.5%, respectively, taking patients as unit. While taking lesions as unit, the sensitivity of AI diagnosis was 55.2%, 81.5%, 92.9% and 79.4% for aneurysms < 3 mm, 3.5 mm, >5 mm, as well as for the total number of aneurysms, respectively; And the accuracy was 53.8%, 75.3%, 86.3% and 72.0%. ②Quantitative analysis: The results of AI and DSA showed good consistency in the measurement of the total aneurysm length diameter, total short diameter, and >5 mm aneurysm length diameter (ICC=0.778, 0.822 and 0.722, P < 0.001); But poor consistency was seen in the measurement for < 3 mm and 3~5 mm aneurysm length diameter (ICC=0.113, 0.142, P>0.05). Conclusion AI has a high value in the application of head and neck CTA in the diagnosis of intracranial aneurysms, with its measurement results to be used as a reference standard in tumors with length diameter >5 mm, it can be further promoted and applied as an auxiliary diagnostic tool.