21-01-2015, 09:55 PM
how to remove skull from brain ?
skull stipping
Posts: 14,118
Threads: 61
Joined: Oct 2014
One of the most common MRI (Magnetic Resonance Imaging) use is a visualization of the brain. The anatomy of the brain is very complicated, therefore, it may be difficult to extract only these structures that have diagnostic value. Consequently, it is so necessary to develop and apply more efficient brain segmentation algorithms. One of the first steps in case of neurological MRI analysis is skull detachment. It is about removing the extra-meningeal tissue from the head image, therefore, it is essential to find the best method to determine the brain and the limits of the skull. In T1-weighted images, cerebrospinal fluid (CSF) space and the skull are dark, so the borders between the brain and skull are well marked, but even strong edges may be unstable due to finite resolution during The acquisition of MRI or the presence of other partial anatomical structures within the brain (connections between the brain and the optic nerves or brainstem). There are many ways to perform this operation, none of them are so large as to constitute a standard procedure. In many cases, there are limitations associated with the development environment, licensing and image entry that make it difficult to remove the skull without specialized software. The proposed method is free of these limitations. It is based on the application of morphological operations and image filtration to improve the result of edge detection and to provide better tissue separation.
The first step in many MRI analysis sequences is the removal of extra-meningeal tissues from the MRI of the entire head. The Brain Surface Extractor (BSE) works by using a Marr-Hildreth edge detector to find a boundary between the brain and the skull. Since the space of the CSF and the skull are dark in a T1 scan, the borders between the brain and the scalp are relatively well defined. However, these strong edges are often interrupted due to the finite resolution of MRI acquisition and actual anatomical connections between the brain and the rest of the body, Optic nerves, brainstem, etc. Thus, BSE uses mathematical morphological operators (erosion, dilation, etc.) to further improve the edge detection result and provide better tissue separation. Additionally, the edge detection result is improved by the application of an anisotropic diffusion filter prior to performing edge detection.