Abstract
In this paper, we present a new automatic robust algorithm to segment multimodal brain MR images with Multiple Sclerosis (MS) lesions. The method performs tissue classification using a Hidden Markov Chain (HMC) model and detects MS lesions as outliers to the model. For this aim, we use the Trimmed Likelihood Estimator (TLE) to extract outliers. Furthermore, neighborhood information is included using the HMC model and we propose to incorporate a priori information brought by a probabilistic atlas.
Keywords
Source Code and Data
No source code files available for this publication.
Reviews
Martin Styner
Tuesday 29 July 2008
The paper employs a hidden markov chain based tissue model and a trimmed likelihood classifier within an EM loop to segment the MS lesions as outliers to the classification.
It is not clear whether the training datasets were used at all. Were they used for the estimation of the best parameter settings such as the trimming parameter h?
Nice paper with very good results, but could use an additional revision to improve the writing. Some sentences should be reformulated, as they are somewhat convoluted.
Simon Warfield
Friday 25 July 2008
This paper develops and applies a new approach to segment MRI images of MS patients.
Particularly interesting contributions are the use of a Hilbert-Peano curve to enable the use of a 1D Markov Chain model, reducing computational complexity compared to a Markov Random Field model, and the use of a trimmed likelihood estimator to reject outliers in the process of estimating a Gaussian mixture model likelihood function.
The authors note that in addition to the lesion class, CSF in particular, but also other tissue classes, may have outlier voxels.
Lesion size criteria and an atlas constraint were utilized to reduce misclassifications.