Changes in myelin integrity are key manifestations of many neurological diseases including multiple sclerosis but precise measurement of myelin in vivo is challenging. The goal of this study was to evaluate myelin content in histological images obtained from a lysolecithin mouse model of demyelination, using a new quantitative method named structure tensor analysis. Injury was targeted at the dorsal column of mice spinal cords. We obtained 16 histological images stained with luxol fast blue for myelin from 9 mice: 9 images from lesion epicenter and 7 from a distant area 500-mu m away from the epicenter. In each image, we categorized 3 tissue types: healthy, completely demyelinated, and partially demyelinated. Structure tensor analysis was applied to quantify the coherency (anisotropy), energy (trace of dominant directions), and angular entropy (degree of disorder) of each tissue. We found that completely demyelinated lesions had significantly lower coherency and energy but higher angular entropy than partially demyelinated and healthy tissues at both the epicenter and distant areas of the injury. In addition, the coherency of healthy tissue was greater than partially demyelinated tissue at each site. Within tissue category, we did not find differences in any measure between spinal cord locations. Our findings suggest that greater myelin integrity is associated with better tissue anisotropy, independent of injury location. Structure tensor analysis may serve as a new tool for quantitative measurement of myelin content in white matter, and this may help understand disease mechanisms and development in MS and other demyelinating disorders. (C) 2014 Elsevier Inc. All rights reserved.