Automatic analysis of trabecular bone structure from knee MRI

Authors:
Joselene Marques;Rabia Granlund;Martin Lillholm;Paola C. Pettersen;Erik B. Dam
Affiliations:
University of Copenhagen, 2100 Copenhagen, Denmark and Biomediq, Copenhagen, Denmark;University of Copenhagen, 2100 Copenhagen, Denmark and Biomediq, Copenhagen, Denmark;Biomediq, Copenhagen, Denmark;Center for Clinical and Basic Research (CCBR), Ballerup, Denmark;Biomediq, Copenhagen, Denmark
Venue:
Computers in Biology and Medicine
Year:
2012

Citing 8
Cited 0

Texture analysis

Handbook of pattern recognition & computer vision
Floating search methods in feature selection

Pattern Recognition Letters
The Gaussian scale-space paradigm and the multiscale local jet

International Journal of Computer Vision
Feature Selection: Evaluation, Application, and Small Sample Performance

IEEE Transactions on Pattern Analysis and Machine Intelligence
Adaptive floating search methods in feature selection

Pattern Recognition Letters - Special issue on pattern recognition in practice VI
Deformable M-Reps for 3D Medical Image Segmentation

International Journal of Computer Vision - Special Issue on Research at the University of North Carolina Medical Image Display Analysis Group (MIDAG)
Feature Extraction: Foundations and Applications (Studies in Fuzziness and Soft Computing)

Feature Extraction: Foundations and Applications (Studies in Fuzziness and Soft Computing)
Weighting of the k-Nearest-Neighbors

ICPR '10 Proceedings of the 2010 20th International Conference on Pattern Recognition

Quantified Score

Hi-index	0.00

Visualization

Abstract

We investigated the feasibility of quantifying osteoarthritis (OA) by analysis of the trabecular bone structure in low-field knee MRI. Generic texture features were extracted from the images and subsequently selected by sequential floating forward selection (SFFS), following a fully automatic, uncommitted machine-learning based framework. Six different classifiers were evaluated in cross-validation schemes and the results showed that the presence of OA can be quantified by a bone structure marker. The performance of the developed marker reached a generalization area-under-the-ROC (AUC) of 0.82, which is higher than the established cartilage markers known to relate to the OA diagnosis.