Review of Aneurysms Detection Methods Focusing on Selected YOLO-Based Models.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Vascular Surgery

Background:

• Aneurysms are dangerous vascular conditions necessitating prompt and precise detection to avert fatalities.
• Early identification of aneurysms is crucial for effective treatment and improved patient outcomes.

Purpose of the Study:

• To review and analyze deep learning approaches, particularly YOLO (You Only Look Once) models, for automated aneurysm detection.
• To evaluate the performance and applicability of different YOLO variants in identifying aneurysms from medical images.

Main Methods:

• A comprehensive review of existing YOLO-based studies for aneurysm detection.
• Practical testing using an annotated dataset of 1342 angiograms, with models trained on 1074 images, validated on 107, and evaluated on 268.
• Comparison of YOLO variants based on architectural characteristics, performance metrics, and clinical relevance.

Main Results:

• Analysis of strengths, limitations, dataset usage, and performance metrics of various YOLO-based studies.
• Demonstration of the comparative performance of selected YOLO variants on an annotated angiogram dataset.
• Identification of key factors influencing the accuracy and robustness of automated aneurysm detection.

Conclusions:

• Deep learning models, especially YOLO variants, hold significant potential for enhancing the accuracy and reliability of aneurysm detection.
• The integration of these detection models can lead to more robust computer-aided diagnostic systems for vascular conditions.
• Further validation and development are needed to establish these systems for widespread clinical use.