An Integrative Classification Model for Multiple Sclerosis Lesion Detection in Multimodal MRI

• Activity Number: 253
• Type: Contributed
• Date/Time: Monday, August 1, 2016 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Imaging
• Author(s): Fengqing Zhang* and Wenxin Jiang and Ji-Ping Wang
• Companies: Drexel University and Northwestern University and Northwestern University
• Keywords: variable selection ; classification ; high dimensional learning ; MRI ; Bayesian modeling ; segmentation

Abstract

We study a classification problem of multiple sclerosis (MS) lesions in three dimensional brain magnetic resonance (MR) images. The number and volume of lesions have been used to evaluate MS disease burden, to track the disease progression, and to evaluate the treatment efficacy. Accurate identification of MS lesions in MR images is challenging due to variability in lesion location, size and shape in addition to anatomical variability between subjects. We propose a supervised classification algorithm for segmenting MS lesions, which integrates the intensity information from multiple MRI modalities, the texture information, and the spatial information in a Bayesian framework. A multinomial logistic regression is used to learn the posterior probability distributions from the intensity information, combined from three MRI modalities. Texture features are selected by the regularized regression model. In addition, the spatial information is included using a Markov random field prior. Results from both the synthetic data and the clinical data have demonstrated the effectiveness of our proposed model for lesion segmentation.