Brain tissues extraction based on improved Brain Extraction Tool algorithm
Jiaqing Qiu, Wenqiang Cheng · 2016
The BET (Brain Extraction Tool) is a useful tool applying on segmenting brain tissue from magnetic resonance image (MRI). It uses deformable model which evolves to fit brain surface by a set of model forces. This method is fast and has a high segmentation precision. However, it has a poor segmentation accuracy of brain parenchyma with lesions. According to the existing problems of the BET algorithm, an improved BET algorithm has been proposed in which we modify the unreasonable parameters such as the force which makes the contour of edge smooth and the force which evolve the edge of brain according to the intensity and gradient information of the MRI images. After that, the new improved algorithm is used to segment not only the normal brain parenchyma but also these with lesions. The improved algorithm is applied to the middle tier of the sequence images and use this method we can get more accurate segmentation results. The similarity of the improved BET algorithm segmentation and the result of artificial reached to more than 90%. Compared with the original algorithm, the improved algorithm is more robust and more accurate.