Automatic Segmentation of Abdominal Adipose Tissue in MRI |
Thomas Hammarshaimb Mosbech, Kasper Arcanius Pilgaard, Allan Vaag, Rasmus Larsen
|
| Abstract | This paper presents a method for automatically segmenting abdominal adipose tissue from 3-dimensional magnetic resonance images. We distinguish between three types of adipose tissue; visceral, deep subcutaneous and superficial subcutaneous. Images are pre-processed to remove the bias field effect of intensity in-homogeneities. This effect is estimated by a thin plate spline extended to fit two classes of automatically sampled intensity points in 3D. Adipose tissue pixels are labelled with fuzzy c-means clustering and locally determined thresholds. The visceral and subcutaneous adipose tissue are separated using deformable models, incorporating information from the clustering. The subcutaneous adipose tissue is subdivided into a deep and superficial part by means of dynamic programming applied to a spatial transformation of the image data. Regression analysis shows good correspondences between our results and total abdominal adipose tissue percentages assessed by dualemission X-ray absorptiometry (R2 = 0.86). |
| Type | Conference paper [With referee] |
| Conference | Proceedings on the Scandinavian Conference on Image Analysis, Ystad, Sweden |
| Year | 2011 pp. 501-511 |
| Publisher | Springer |
| Series | Lecture Notes in Computer Science |
| Note | DOI 10.1007/978-3-642-21227-7 |
| BibTeX data | [bibtex] |
| IMM Group(s) | Image Analysis & Computer Graphics |