Abstract | ||
---|---|---|
In this paper we propose a method for segmenting structures in Magnetic Resonance Images (MRI) using the well known kernel Adatron algorithm. The method allows the segmentation in feature space of 2D structures present in each slice of a set of MRI data. The resulting 2D segmented regions are used as input of a kernel SVM algorithm to produce a three dimensional model of the object. This procedure enables the physician the visualization of the complete 'real' structure, difficult to perceive from the individual 2D slices of the original MRI data. The representation of the 3D model is obtained as an implicit function, allowing volume, distances and related measurements. The methodology is applied to the segmentation and modeling of a specific structure: a brain tumor from a real data set of a human brain. |
Year | DOI | Venue |
---|---|---|
2004 | 10.1007/978-3-540-30498-2_64 | ADVANCES IN ARTIFICIAL INTELLIGENCE - IBERAMIA 2004 |
Keywords | Field | DocType |
feature space,magnetic resonance image,three dimensional | Kernel (linear algebra),Feature vector,Pattern recognition,Visualization,Segmentation,Computer science,Support vector machine,Implicit function,Artificial intelligence,Kernel method,Sequential minimal optimization,Distributed computing | Conference |
Volume | ISSN | Citations |
3315 | 0302-9743 | 10 |
PageRank | References | Authors |
1.20 | 5 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Cristina García | 1 | 19 | 3.56 |
José Alí Moreno | 2 | 65 | 8.60 |