Share Email Print
cover

Proceedings Paper

A learning-based automatic spinal MRI segmentation
Author(s): Xiaoqing Liu; Jagath Samarabandu; Greg Garvin; Rethy Chhem; Shuo Li
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

Image segmentation plays an important role in medical image analysis and visualization since it greatly enhances the clinical diagnosis. Although many algorithms have been proposed, it is still challenging to achieve an automatic clinical segmentation which requires speed and robustness. Automatically segmenting the vertebral column in Magnetic Resonance Imaging (MRI) image is extremely challenging as variations in soft tissue contrast and radio-frequency (RF) in-homogeneities cause image intensity variations. Moveover, little work has been done in this area. We proposed a generic slice-independent, learning-based method to automatically segment the vertebrae in spinal MRI images. A main feature of our contributions is that the proposed method is able to segment multiple images of different slices simultaneously. Our proposed method also has the potential to be imaging modality independent as it is not specific to a particular imaging modality. The proposed method consists of two stages: candidate generation and verification. The candidate generation stage is aimed at obtaining the segmentation through the energy minimization. In this stage, images are first partitioned into a number of image regions. Then, Support Vector Machines (SVM) is applied on those pre-partitioned image regions to obtain the class conditional distributions, which are then fed into an energy function and optimized with the graph-cut algorithm. The verification stage applies domain knowledge to verify the segmented candidates and reject unsuitable ones. Experimental results show that the proposed method is very efficient and robust with respect to image slices.

Paper Details

Date Published: 11 March 2008
PDF: 8 pages
Proc. SPIE 6914, Medical Imaging 2008: Image Processing, 69143L (11 March 2008); doi: 10.1117/12.769891
Show Author Affiliations
Xiaoqing Liu, The Univ. of Western Ontario (Canada)
Jagath Samarabandu, The Univ. of Western Ontario (Canada)
Greg Garvin, St. Joseph's Health Care (Canada)
Rethy Chhem, London Health Sciences Ctr. (Canada)
Shuo Li, GE Healthcare (Canada)


Published in SPIE Proceedings Vol. 6914:
Medical Imaging 2008: Image Processing
Joseph M. Reinhardt; Josien P. W. Pluim, Editor(s)

© SPIE. Terms of Use
Back to Top