الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 92 |  |  |
| | | from | 92 | | |
Abstract
The segmentation of multiple sclerosis (MS) lesions play a core role in diagnosis and monitoring MS. Magnetic resonance imaging (MRI) is the most common method to evaluate MS lesions. Despite the manual segmentation is the standard in clinical practice, it is unproductive with a large amount of data that restricts the usage of accurate quantitative measurement in clinical practice. Therefore, the need for effective automated segmentation techniques is critical. However, significant spatial variability between the structure of brain lesions makes it more challenges. Recently, convolutional neural networks (CNN), in particular, the region-based CNN (R- CNN) have attained tremendous progress in object recognition because of its ability to learn and represent features. CNN has gained popularity in brain imaging, especially in tissue and brain segmentation. In this thesis, an automated technique for MS lesion segmentation is proposed, which based on a 3D patch-wise R-CNN. The proposed method includes two stages: firstly, segmenting MS lesions in T1-w, T2-w and FLAIR sequences using R-CNN. Secondly, an adaptive neuro-fuzzy inference system (ANFIS) is applied to fuse the results of T2-W and FLAIR modalities. To evaluate the performance of the proposed method, the public MICCAI2008 MS challenge dataset is employed to segment MS lesion. The experimental results show the good performance of the proposed method compared to the state-of-the-art MS lesion segmentation.