Automated Computerized Method for the Detection of Unruptured Cerebral Aneurysms in DSA Images

Introduction: We present in this paper an automatic detection system of brain aneurysms in Digital Subtraction Angiography (DSA) images. Methods: The approach proposed here relies on two main steps: A region-based segmentation using the Otsu method is first employed in order to extract the vascular structure. Then a combination of the Zernike moments and MSER detector is used to detect aneurysms. Result: In our experiments, the evaluation results over 14 DSA images have given very satisfactory results. Conclusion: The achieved sensitivity of the proposed cerebral aneurysm detection method was 100% with 0.78 false positive rate.