Brain Image Vessel Skeletonization Analysis Using Deep Learning Salient Vascular Candidate Extraction Method

Abstract

Image processing plays a crucial role in modern medical imaging, enhancing the ability to analyze and interpret complex medical data. By applying advanced algorithms to medical images such as X-rays, MRIs, CT scans, and ultrasounds, image processing techniques allow for the extraction of valuable diagnostic information, improving the accuracy and efficiency of disease detection and treatment planning. Key methods in medical image processing include image enhancement, segmentation, registration, and feature extraction, which assist clinicians in identifying anomalies such as tumors, lesions, and organ abnormalities. Recent advancements in deep learning have markedly enhanced the accuracy and efficiency of brain image analysis. By making it possible to analyze complicated neuroimaging data more accurately and efficiently, the use of deep learning techniques in brain image processing has greatly advanced the area of neuro science. For the clinical evaluation of intracranial vascular disorders, brain blood vessel extraction is a crucial concern. This work formulates the problem of vessel extraction as a connected region classification problem. A post-processing step is added to the image processing process to collect salient vessel candidate extraction method (SVCEM), and an enhanced multi-scale filtering method is used to increase vessel connection. A neural network classifier is then trained using the features that are computed after SVCEM is broken down to connected regions. The trained neural network is used to examine each connected region separately for extraction, taking to the values of nearby voxels that are part of the connection. The extraction results demonstrate the validity of the proposed approach.