An Early Warning Model of Unfavorable Outcomes Following Endovascular Interventional Treatment of Intracranial Aneurysm

An Early Warning Model of Unfavorable Outcomes Following Endovascular Interventional Treatment of Intracranial Aneurysms Based on Medical Image Analysis and Deep Learning Algorithm

Endovascular treatment has become one of the primary treatment methods for intracranial aneurysms. The unfavorable outcomes during follow-up included aneurysm recurrence and long-term incomplete-occlusion, which would bring a high risk of rebleeding and retreatment. Previous studies have tried to predict the outcomes of aneurysms following endovascular treatment based on aneurysm characteristics including morphology, embolization packing degree, etc, but the conclusion was inconsistent. Hemodynamics of aneurysms and parent artery played a greater role in predicting outcomes following endovascular treatments. Investigators also found that the outcomes were determined by many factors, in which the demography, clinical indicators, treatment methods, and material selection can not be ignored, and the mechanism of unfavorable imaging outcomes should be explored using large samples of clinical cases and numerous variable parameters. The pre-experiment of investigators confirmed that artificial intelligence technology can meet the calculation requirements for deep mining and analysis of large sample data. This study aims to use the deep learning model to identify relevant risk factors and weights, establish a stable and accurate prediction model, then incorporate the prospective study to verify the model. The results will be very helpful in accurately predicting the adverse outcomes such as recurrence and long-term non-occlusion after endovascular treatment and help to improve the therapeutic strategy and avoid risk factors. Besides, the occurrence of ischemic or hemorrhagic complications during follow-up may affect the final follow-up outcome, so the analysis was included as one of the outcome events to evaluate the prognosis after intervention.

Inclusion Criteria:

• Clinical diagnosis of intracranial aneurysm；
• Endovascular intervention was performed；
• The age is more than 18 years and less than 80 years;
• At least one follow-up of imaging data must be digital subtraction angiography with a time interval of 12 months or more;
• The quality of image data can satisfy morphological measurement and hemodynamic calculation;
• The patient family members were informed and consented to participate in the study.

Exclusion Criteria:

• Dynamic aneurysms with cerebrovascular malformation;
• Dissecting, fusiform or thrombotic aneurysms;
• Follow-up images were not digital subtraction angiography or data quality can not meet the hydrodynamic analysis.