Automatic Segmentation of Putamen using Geometric Moment Invariants

We propose an automatic segmentation of Magnetic Resonance (MR) images of brain structures in 3D-space, ocusing on the putamen. Our task is to classify voxels into two classes - being inside or outside the putamen. We propose the use of a new feature extraction method for brain structure segmentation – geometric moment invariants (GMI). GMIs are eleven moment invariants calculated from first-order, second-order and third-order 3D regular moments that reflect underlying anatomical structures of the brain. Experimental results, using GMIs as the features, along with spatial information to feed an artificial neural network, show that these features improve the segmentation performance. The proposed representation method can be applied to other brain structures such as the hippocampus, the amygdale, and the caudate.

Automatic Segmentation of Putamen using Geometric Moment Invariants Keywords:

Putamen , Geometric Moment Invariants , Artificial Neural Network