A New Image Segmentation Method Based on Three-Dimensional Neural Network
2012 ◽
Vol 490-495
◽
pp. 157-161
Keyword(s):
In this paper, a three-dimensional probabilistic approach for MR brain image segmentation is proposed. Based on the noise-free representative reference vectors provided by SOM, the results of the 3D-PNN method are superior to other traditional algorithms. In addition to the 3D-PNN architecture, a fast three-step training method is proposed. The proposed approach also incorporates structure tensor to find appropriate feature sets for the 3D-PNN with respect to resulting classification accuracy. Computational results with simulated MR brain images have shown the promising performance of the proposed method.
Keyword(s):
1995 ◽
Vol 42
(11)
◽
pp. 1069-1078
◽
Keyword(s):
Keyword(s):
2017 ◽
Vol 62
(6)
◽
pp. 581-590
◽
Keyword(s):
2005 ◽
pp. 505-513
◽
2013 ◽
Vol 67
(16)
◽
pp. 18-20
Keyword(s):