Evaluating Deep Learning Algorithms for Aneurysm Detection Using CNN and VGG-16 with Random Forest Integration

Aneurysms are dangerous for health, especially in cases when the disease's fatality threatens the human brain. That's why the accuracy of aneurysm detections is vital. This paper aims to analyze the efficiency of various algorithms in Deep Learning in order to improve the accuracy of aneurysm detection from CTA images. In more detail, it has been done using three aspects in the experiment, which are Convolutional Neural Networks (CNN), Visual Geometry Group 16 (VGG-16), and Random Forest (RF), and VGG-16 combined with Random Forest (RF) when diagnosing aneurysms. This study creates models using the algorithms utilized in the experiments done. To assess the performance of the models, a dataset, which was composed of CTA images was used and metrics whereby accuracy, precision, recall, and F1 score were used. It has been found that the model that combines CNN with RF provided the highest accuracy of 98.04%, precision of 98.61%, and F1-score of 97.7%. The results show that the proposed RF with CNN can enhance the aneurysm detection and point out specific recommendations for future studies, such as handling the sample diversification and regarding the model interpretability for clinical use.