Automated Method for Intracranial Aneurysm Classification Using Deep Learning.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neurosurgery

Background:

• Intracranial aneurysms (IA) are a significant cause of subarachnoid hemorrhage, often leading to severe disability or death.
• Accurate and timely diagnosis of IA is crucial for effective patient management and improved outcomes.
• Current diagnostic methods can be time-consuming, highlighting the need for automated detection systems.

Purpose of the Study:

• To develop and evaluate a novel 2D Convolutional Neural Network (CNN) for automated detection of intracranial aneurysms (IA).
• To compare the performance of the proposed CNN against established networks like ResNet and VGG using a public dataset.
• To assess the efficiency (size and speed) of the proposed network for potential clinical application.

Main Methods:

• A 2D Convolutional Neural Network (CNN) was designed and trained for IA detection.
• The proposed CNN was evaluated on a publicly available dataset of 611 images.
• Performance was compared with ResNet (50, 101, 152) and VGG networks, focusing on accuracy, network size, and classification time.

Main Results:

• The proposed CNN achieved an overall accuracy of 98% in detecting intracranial aneurysms.
• ResNet 152 demonstrated higher accuracy but resulted in a significantly larger model and longer classification times.
• The proposed network was substantially smaller and faster than ResNet 152, while VGG network performance was inadequate (20% accuracy).

Conclusions:

• The developed 2D CNN presents a highly accurate and efficient automated system for intracranial aneurysm detection.
• This AI-driven approach has the potential to significantly aid physicians in diagnosing IA, improving patient care.
• The findings support the integration of this optimized CNN into clinical workflows for faster and more reliable IA screening.