Name
Abstract | ||
|---|---|---|
Many modern forms of segmentation and registration require manual input making it a tedious and time-consuming process. There have been some successes with automating these methods, but these tend to be unreliable because of inherent variations in anatomical shapes and image quality. It is toward this goal that we have developed an automated method of generating landmarks for registration that will not require supervision or manual initialization. We have chosen medial based image features because they have proven robust against image noise and shape variation, and provide the rotationally invariant properties of dimensionality and scale, which can be used by a unary metric. We introduce a new metric for comparing the geometric relationships between medial features, which overcomes problems introduced by symmetry within a medial feature. With these metrics, we are able to find correspondences between pairs and triplets of features in the two images. We demonstrate these methods on three different datasets. It is envisioned that this system will become the basis for generating medial node models that can be registered between two images. |
Year | DOI | Venue |
|---|---|---|
2003 | 10.1007/978-3-540-39701-4_36 | BIOMEDICAL IMAGE REGISTRATION |
Keywords | Field | DocType |
image quality,image features | Computer vision,Unary operation,Pattern recognition,Feature (computer vision),Computer science,Segmentation,Medial axis,Image quality,Image noise,Invariant (mathematics),Artificial intelligence,Initialization | Conference |
Volume | ISSN | Citations |
2717 | 0302-9743 | 0 |
PageRank | References | Authors |
0.34 | 7 | 4 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Robert J. Tamburo | 1 | 38 | 3.59 |
| Aaron Cois | 2 | 10 | 1.75 |
| Damion Shelton | 3 | 45 | 4.93 |
| George D. Stetten | 4 | 146 | 22.70 |