Efficient quantification of intracranial aneurysms is paramount for effective treatment and management. However, variability among experts and the complexity of anatomical locations, especially with small aneurysms, pose significant challenges for radiologists. In this study, we propose a novel pre-processing method that is context-aware and topology-preserving, aimed at accurately reducing the volume of interest in Magnetic Resonance Angiography (MRA) images. This method ensures that the anatomical integrity and relevant topological features of the aneurysms are maintained throughout the pre-processing step. The study includes both pre- and post-treatment analyses of intracranial aneurysms from public and private databases. Furthermore, with the help of radiologists, annotations of major cerebral vessels were created for both databases, which is a preliminary task for intracranial aneurysm treatment and follow-up studies. The methodology has been validated through extensive experimentation, demonstrating its efficacy in preserving critical topological characteristics while significantly enhancing the quantification process. Our findings suggest that this morphology-based pre-processing approach is promising for improving the severe class imbalance in intracranial aneurysm segmentation, indicating its potential utility in clinical neuroradiology.