Loading

3DMSNET: 3D CNN Based Brain MRI Segmentation
Junaid Ahmad1, Bhanu Bhaskar2, Haresh Seetharaman3, Ajay Kumar4, J. Arunnehru5
1Junaid Ahmad, Department of Computer Science, SRM Institute of Science & Technology, Chennai (Tamil Nadu), India.
2Bhanu Bhaskar Kotagiri, Department of Computer Science, SRM Institute of Science & Technology, Chennai (Tamil Nadu), India.
3Haresh Seetharaman, Department of Computer Science, SRM Institute of Science & Technology, Chennai (Tamil Nadu), India.
4Ajay Kumar, Department of Computer Science, SRM Institute of Science & Technology, Chennai (Tamil Nadu), India.
5J.Arunnehru, Department of Computer Science, SRM Institute of Science & Technology, Chennai (Tamil Nadu), India.
Manuscript received on 25 August 2019 | Revised Manuscript received on 01 September 2019 | Manuscript Published on 14 September 2019 | PP: 107-110 | Volume-8 Issue-5S3, July 2019 | Retrieval Number: E10270785S319/19©BEIESP | DOI: 10.35940/ijeat.E1027.0785S319
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Segmentation of the brain images has become an important task to analyze the abnormality in infants. Automatic methods are important as the infant brain growth has to be tracked and it is almost impossible for an individual to manually segment the MRI data on particular intervals. The manual segmentation tasks are time-consuming and require highly skilled professionals to segment images. Automatic segmentation methods have gained huge support for segmenting MRI images. Several segmentation methods lack accuracies due to nearest neighbor or self-similarity problems. The CNNs have outperformed the traditional methods and are proving to be more reliable day by day. The proposed method is a patch-based method which uses 3DMSnet (3D Multi-Scale Network) for segmentation. The model is evaluated on BrainWeb and other publicly available datasets.
Keywords: Segmentation, Nearest-Neighbor, Self-Similarity, Patch-Based, Brainweb, CNN, MRI, 3DMSnet.
Scope of the Article: 3D Printing