Diffusion Tensor Imaging Segmentation Research Articles

Brain multi-contrast, multi-atlas segmentation of diffusion tensor imaging and ensemble learning automatically diagnose late-life depression

We investigated the potential of machine learning for diagnostic classification in late-life major depression based on an advanced whole brain white matter segmentation framework. Twenty-six late-life depression and 12 never depressed individuals aged > 55 years, matched for age, MMSE, and education underwent brain diffusion tensor imaging and a multi-contrast, multi-atlas segmentation in MRIcloud. Fractional anisotropy volume, mean fractional anisotropy, trace, axial and radial diffusivity (RD) extracted from 146 white matter parcels for each subject were used to train and test the AdaBoost classifier using stratified 12-fold cross validation. Performance was evaluated using various measures. The statistical power of the classifier was assessed using label permutation test. Statistical analysis did not yield significant differences in DTI measures between the groups. The classifier achieved a balanced accuracy of 71% and an Area Under the Receiver Operator Characteristic Curve (ROC-AUC) of 0.81 by trace, and a balanced accuracy of 70% and a ROC-AUC of 0.80 by RD, in limbic, cortico-basal ganglia-thalamo-cortical loop, brainstem, external and internal capsules, callosal and cerebellar structures. Both indices shared important structures for classification, while fornix was the most important structure for classification by both indices. The classifier proved statistically significant, as trace and RD ROC-AUC scores after permutation were lower than those obtained with the actual data (P = 0.022 and P = 0.024, respectively). Similar results were obtained with the Gradient Boosting classifier, whereas the RBF-kernel Support Vector Machine with k-best feature selection did not exceed the chance level. Finally, AdaBoost significantly predicted the class using all features together. Limitations are discussed. The results encourage further investigation of the implemented methods for computer aided diagnostics and anatomically informed therapeutics.

Determining the OPTIMAL DTI analysis method for application in cerebral small vessel disease

BackgroundDTI is sensitive to white matter (WM) microstructural damage and has been suggested as a surrogate marker for phase 2 clinical trials in cerebral small vessel disease (SVD). The study’s objective is to establish the best way to analyse the diffusion-weighted imaging data in SVD for this purpose. The ideal method would be sensitive to change and predict dementia conversion, but also straightforward to implement and ideally automated. As part of the OPTIMAL collaboration, we evaluated five different DTI analysis strategies across six different cohorts with differing SVD severity. MethodsThose 5 strategies were: (1) conventional mean diffusivity WM histogram measure (MD median), (2) a principal component-derived measure based on conventional WM histogram measures (PC1), (3) peak width skeletonized mean diffusivity (PSMD), (4) diffusion tensor image segmentation θ (DSEG θ) and (5) a WM measure of global network efficiency (Geff). The association between each measure and cognitive function was tested using a linear regression model adjusted by clinical markers. Changes in the imaging measures over time were determined. In three cohort studies, repeated imaging data together with data on incident dementia were available. The association between the baseline measure, change measure and incident dementia conversion was examined using Cox proportional-hazard regression or logistic regression models. Sample size estimates for a hypothetical clinical trial were furthermore computed for each DTI analysis strategy. ResultsThere was a consistent cross-sectional association between the imaging measures and impaired cognitive function across all cohorts. All baseline measures predicted dementia conversion in severe SVD. In mild SVD, PC1, PSMD and Geff predicted dementia conversion. In MCI, all markers except Geff predicted dementia conversion. Baseline DTI was significantly different in patients converting to vascular dementia than to Alzheimer’ s disease. Significant change in all measures was associated with dementia conversion in severe but not in mild SVD. The automatic and semi-automatic measures PSMD and DSEG θ required the lowest minimum sample sizes for a hypothetical clinical trial in single-centre sporadic SVD cohorts. ConclusionDTI parameters obtained from all analysis methods predicted dementia, and there was no clear winner amongst the different analysis strategies. The fully automated analysis provided by PSMD offers advantages particularly for large datasets.

Structural connectivity-based segmentation of the human entorhinal cortex

The medial (MEC) and lateral entorhinal cortex (LEC), widely studied in rodents, are well defined and characterized. In humans, however, the exact locations of their homologues remain uncertain. Previous functional magnetic resonance imaging (fMRI) studies have subdivided the human EC into posteromedial (pmEC) and anterolateral (alEC) parts, but uncertainty remains about the choice of imaging modality and seed regions, in particular in light of a substantial revision of the classical model of EC connectivity based on novel insights from rodent anatomy. Here, we used structural, not functional imaging, namely diffusion tensor imaging (DTI) and probabilistic tractography to segment the human EC based on differential connectivity to other brain regions known to project selectively to MEC or LEC. We defined MEC as more strongly connected with presubiculum and retrosplenial cortex (RSC), and LEC as more strongly connected with distal CA1 and proximal subiculum (dCA1pSub) and lateral orbitofrontal cortex (OFC). Although our DTI segmentation had a larger medial-lateral component than in the previous fMRI studies, our results show that the human MEC and LEC homologues have a border oriented both towards the posterior-anterior and medial-lateral axes, supporting the differentiation between pmEC and alEC.

Diffusion Tensor Imaging Correlates of Concussion Related Cognitive Impairment.

Introduction: Cognitive impairment after concussion has been widely reported, but there is no reliable imaging biomarker that predicts the severity of cognitive decline post-concussion. This study tests the hypothesis that patients with a history of concussion and persistent cognitive impairment have fractional anisotropy (FA) and mean diffusivity (MD) values from diffusion tensor imaging (DTI) that are specifically associated with poor performance on the Montreal Cognitive Assessment (MoCA).Methods: Fifty-three subjects (19 females) with concussions and persistent cognitive symptoms had MR imaging and the MoCA. Imaging was analyzed by atlas-based, whole-brain DTI segmentation and FLAIR lesion segmentation. Then, we conducted a random forest-based recursive feature elimination (RFE) with 10-fold cross-validation on the entire dataset, and with partial correlation adjustment for age and lesion load.Results: RFE showed that 11 DTI variables were found to be important predictors of MoCA scores. Partial correlation analyses, corrected for age and lesion load, showed significant correlations between MoCA scores and right fronto-temporal regions: inferior temporal gyrus MD (r = −0.62, p = 0.00001), middle temporal gyrus MD (r = −0.54, p = 0.0001), angular gyrus MD (r = −0.48, p = 0.0008), and inferior frontal gyrus FA (r = 0.44, p = 0.002).Discussion: This is the first study to demonstrate a correlation between MoCA scores and DTI variables in patients with a history of concussion and persistent cognitive impairment. This kind of research will significantly increase our understanding of why certain persons have persistent cognitive changes after concussion which, in turn, may allow us to predict persistent impairment after concussion and suggest new interventions.

Predicting Dementia in Cerebral Small Vessel Disease Using an Automatic Diffusion Tensor Image Segmentation Technique.

Background and Purpose- Cerebral small vessel disease (SVD) is the most common cause of vascular cognitive impairment, with a significant proportion of cases going on to develop dementia. We explore the extent to which diffusion tensor image segmentation technique (DSEG; which characterizes microstructural damage across the cerebrum) predicts both degree of cognitive decline and conversion to dementia, and hence may provide a useful prognostic procedure. Methods- Ninety-nine SVD patients (aged 43-89 years) underwent annual magnetic resonance imaging scanning (for 3 years) and cognitive assessment (for 5 years). DSEG-θ was used as a whole-cerebrum measure of SVD severity. Dementia diagnosis was based Diagnostic and Statistical Manual of Mental Disorders V criteria. Cox regression identified which DSEG measures and vascular risk factors were related to increased risk of dementia. Linear discriminant analysis was used to classify groups of stable versus subsequent dementia diagnosis individuals. Results- DSEG-θ was significantly related to decline in executive function and global cognition (P<0.001). Eighteen (18.2%) patients converted to dementia. Baseline DSEG-θ predicted dementia with a balanced classification rate=75.95% and area under the receiver operating characteristic curve=0.839. The best classification model included baseline DSEG-θ, change in DSEG-θ, age, sex, and premorbid intelligence quotient (balanced classification rate of 79.65%; area under the receiver operating characteristic curve=0.903). Conclusions- DSEG is a fully automatic technique that provides an accurate method for assessing brain microstructural damage in SVD from a single imaging modality (diffusion tensor imaging). DSEG-θ is an important tool in identifying SVD patients at increased risk of developing dementia and has potential as a clinical marker of SVD severity.

Diffusion Tensor Image segmentation based on multi-atlas Active Shape Model

Active Shape Model (ASM) has been successfully applied in the segmentation of Diffusion Tensor Magnetic Resonance Image (DT-MRI, referred to as DTI) of brain. However, due to multiple anatomical structure types, irregular shapes, small gray-scale and large amount of these images, perfect segmentation performance could not be achieved. Especially, it is sensitive to initial values with high computational complexity. In this paper, we introduce the gray information of multiple atlases and the prior information of target shapes into the ASM and propose the Multi-Atlas Active Shape Model (referred to as MA-ASM) approach for DTI segmentation. It was evaluated in a manually labeled database with 7 Region of Interest (ROI)s for each of 20 subjects. In comparison with the state of art method of STAPLE (Simultaneous Truth Performance Level Estimation), the proposed algorithm was closer to the manual segmentation shape by subjective visual effects, and had higher overlap rates and lower error detection rates on quantitative analysis than STAPLE.

Diffusion tensor image segmentation of the cerebrum provides a single measure of cerebral small vessel disease severity related to cognitive change.

Cerebral small vessel disease (SVD) is the primary cause of vascular cognitive impairment and is associated with decline in executive function (EF) and information processing speed (IPS). Imaging biomarkers are needed that can monitor and identify individuals at risk of severe cognitive decline. Recently there has been interest in combining several magnetic resonance imaging (MRI) markers of SVD into a unitary score to describe disease severity. Here we apply a diffusion tensor image (DTI) segmentation technique (DSEG) to describe SVD related changes in a single unitary score across the whole cerebrum, to investigate its relationship with cognitive change over a three-year period.98 patients (aged 43–89) with SVD underwent annual MRI scanning and cognitive testing for up to three years. DSEG provides a vector of 16 discrete segments describing brain microstructure of healthy and/or damaged tissue. By calculating the scalar product of each DSEG vector in reference to that of a healthy ageing control we generate an angular measure (DSEG θ) describing the patients' brain tissue microstructural similarity to a disease free model of a healthy ageing brain. Conventional MRI markers of SVD brain change were also assessed including white matter hyperintensities, cerebral atrophy, incident lacunes, cerebral-microbleeds, and white matter microstructural damage measured by DTI histogram parameters. The impact of brain change on cognition was explored using linear mixed-effects models. Post-hoc sample size analysis was used to assess the viability of DSEG θ as a tool for clinical trials.Changes in brain structure described by DSEG θ were related to change in EF and IPS (p < 0.001) and remained significant in multivariate models including other MRI markers of SVD as well as age, gender and premorbid IQ. Of the conventional markers, presence of new lacunes was the only marker to remain a significant predictor of change in EF and IPS in the multivariate models (p = 0.002). Change in DSEG θ was also related to change in all other MRI markers (p < 0.017), suggesting it may be used as a surrogate marker of SVD damage across the cerebrum. Sample size estimates indicated that fewer patients would be required to detect treatment effects using DSEG θ compared to conventional MRI and DTI markers of SVD severity.DSEG θ is a powerful tool for characterising subtle brain change in SVD that has a negative impact on cognition and remains a significant predictor of cognitive change when other MRI markers of brain change are accounted for. DSEG provides an automatic segmentation of the whole cerebrum that is sensitive to a range of SVD related structural changes and successfully predicts cognitive change. Power analysis shows DSEG θ has potential as a monitoring tool in clinical trials. As such it may provide a marker of SVD severity from a single imaging modality (i.e. DTIs).

Exploiting an advanced DTI segmentation technique towards differentiation of GBM and MET

Introduction It is commonly accepted that differentiation between glioblastoma multiforme (GBM) and solitary metastases (MET), relying on conventional Magnetic Resonance Imaging (MRI), in daily clinical practice, remains controversial. Purpose In the frame of such differentiation by means of quantitative image analysis of advanced MR techniques, such as Diffusion Tensor Imaging (DTI), the initial step of tumor segmentation is essential. Materials and methods In this study a proposed state-of-the-art segmentation technique was implemented, based on isotropic (p) and anisotropic (q) maps, derived from diffusion tensor decomposition. The unsupervised k-medians clustering of the 2D (p,q) histogram ( k = 16, account for 16 different types of brain tissues) results in whole brain segmented maps, where brain tumor lesions present distinctive boundaries. The technique has been tested on a case sample of 10 GBM and 10 MET patients, who underwent preoperative DTI scans at 3Tesla. Results Initial pilot evaluation of the produced brain color maps, by expert observers, demonstrated a potential role of specific tissue segments in precise determination of tumor’s margins, including intratumoral/peritumoral regions. In addition, due to its automated character, the technique is expected to deal with observer variabilities, introduced by manual ROI sampling of the above mentioned tumor regions, representing the current clinical standard. Conclusion The technique implemented lends itself to 3D tumor modeling and is expected to contribute in GMB and metastases differentiation, by means of 3D surface quantitative descriptors, complemented by 3D whole tumor texture analysis.

Brain tumor classification using the diffusion tensor image segmentation (D-SEG) technique.

BackgroundThere is an increasing demand for noninvasive brain tumor biomarkers to guide surgery and subsequent oncotherapy. We present a novel whole-brain diffusion tensor imaging (DTI) segmentation (D-SEG) to delineate tumor volumes of interest (VOIs) for subsequent classification of tumor type. D-SEG uses isotropic (p) and anisotropic (q) components of the diffusion tensor to segment regions with similar diffusion characteristics.MethodsDTI scans were acquired from 95 patients with low- and high-grade glioma, metastases, and meningioma and from 29 healthy subjects. D-SEG uses k-means clustering of the 2D (p,q) space to generate segments with different isotropic and anisotropic diffusion characteristics.ResultsOur results are visualized using a novel RGB color scheme incorporating p, q and T2-weighted information within each segment. The volumetric contribution of each segment to gray matter, white matter, and cerebrospinal fluid spaces was used to generate healthy tissue D-SEG spectra. Tumor VOIs were extracted using a semiautomated flood-filling technique and D-SEG spectra were computed within the VOI. Classification of tumor type using D-SEG spectra was performed using support vector machines. D-SEG was computationally fast and stable and delineated regions of healthy tissue from tumor and edema. D-SEG spectra were consistent for each tumor type, with constituent diffusion characteristics potentially reflecting regional differences in tissue microstructure. Support vector machines classified tumor type with an overall accuracy of 94.7%, providing better classification than previously reported.ConclusionsD-SEG presents a user-friendly, semiautomated biomarker that may provide a valuable adjunct in noninvasive brain tumor diagnosis and treatment planning.

Multi-channel features based automated segmentation of diffusion tensor imaging using an improved FCM with spatial constraints

The aim of this study is to design an improved FCM with constraints algorithm (iFCM_S) for brain tissue segmentation based on diffusion tensor imaging (DTI). The fuzzy c-means clustering algorithm has been widely used in many medical image segmentations. However, the conventionally standard FCM algorithm is sensitive to noise because of not taking into account the membership function information and the spatial contextual information. To overcome this problem, an improved FCM with spatial constraints algorithm for image segmentation is presented in this paper, which is formulated by modifying the objective function and the membership function of the standard fuzzy c-means algorithm to enhance the noise immunity. In addition, due to multi-channel features of DTI data providing more information to the tissue segmentation comparing to single channel feature, we use the proposed algorithm on the multi-channel features of DTI to implement brain tissue segmentation. The experiments on both simulated images and real-world datasets show that our proposed method is more effective than the conventional segmentation methods.

A statistical approach to segmentation of diffusion tensor imaging

The aim of this study is to design a statistical segmentation technique to allow extraction of grey matter, white matter and cerebral spinal fluid volumes from diffusion tensor imaging. Four channel maps of the DTI are used as the input features, which provide more information for brain tissue segmentation compared with single channel map. An Improved Bayesian decision in the subspace spanned by the eigenvectors which are associated with the smaller eigenvalues in each class is adopted as the brain tissue segmentation criterion. Our method performed well, giving an average segmentation accuracy of about 0.88, 0.85 and 0.76 for white matter, gray matter and cerebrospinal fluid respectively in terms of volume overlap.

Reduced field-of-view DTI segmentation of cervical spine tissue

The number of diffusion tensor imaging (DTI) studies regarding the human spine has considerably increased and it is challenging because of the spine’s small size and artifacts associated with the most commonly used clinical imaging method. A novel segmentation method based on the reduced field-of-view (rFOV) DTI dataset is presented in cervical spinal canal cerebrospinal fluid, spinal cord grey matter and white matter classification in both healthy volunteers and patients with neuromyelitis optica (NMO) and multiple sclerosis (MS). Due to each channel based on high resolution rFOV DTI images providing complementary information on spinal tissue segmentation, we want to choose a different contribution map from multiple channel images. Via principal component analysis (PCA) and a hybrid diffusion filter with a continuous switch applied on fourteen channel features, eigen maps can be obtained and used for tissue segmentation based on the Bayesian discrimination method. Relative to segmentation by a pair of expert readers, all of the automated segmentation results in the experiment fall in the good segmentation area and performed well, giving an average segmentation accuracy of about 0.852 for cervical spinal cord grey matter in terms of volume overlap. Furthermore, this has important applications in defining more accurate human spinal cord tissue maps when fusing structural data with diffusion data. rFOV DTI and the proposed automatic segmentation outperform traditional manual segmentation methods in classifying MR cervical spinal images and might be potentially helpful for detecting cervical spine diseases in NMO and MS.

Brain tissue classification based on DTI using an improved Fuzzy C-means algorithm with spatial constraints

We present an effective method for brain tissue classification based on diffusion tensor imaging (DTI) data. The method accounts for two main DTI segmentation obstacles: random noise and magnetic field inhomogeneities. In the proposed method, DTI parametric maps were used to resolve intensity inhomogeneities of brain tissue segmentation because they could provide complementary information for tissues and define accurate tissue maps. An improved fuzzy c-means with spatial constraints proposal was used to enhance the noise and artifact robustness of DTI segmentation. Fuzzy c-means clustering with spatial constraints (FCM_S) could effectively segment images corrupted by noise, outliers, and other imaging artifacts. Its effectiveness contributes not only to the introduction of fuzziness for belongingness of each pixel but also to the exploitation of spatial contextual information. We proposed an improved FCM_S applied on DTI parametric maps, which explores the mean and covariance of the feature spatial information for automated segmentation of DTI. The experiments on synthetic images and real-world datasets showed that our proposed algorithms, especially with new spatial constraints, were more effective.

DTI segmentation via the combined analysis of connectivity maps and tensor distances

We describe a novel approach to extract the neural tracts of interest from a diffusion tensor image (DTI). Compared to standard streamline tractography, existing probabilistic methods are able to capture fiber paths that deviate from the main tensor diffusion directions. At the same time, tensor clustering methods are able to more precisely delimit the border of the bundle. To the best of our knowledge, we propose the first algorithm which combines the advantages supplied by probabilistic and tensor clustering approaches. The algorithm includes a post-processing step to limit partial-volume related segmentation errors. We extensively test the accuracy of our algorithm on different configurations of a DTI software phantom for which we systematically vary the image noise, the number of gradients, the geometry of the fiber paths and the angle between adjacent and crossing fiber bundles. The reproducibility of the algorithm is supported by the segmentation of the corticospinal tract of nine patients. Additional segmentations of the corticospinal tract, the arcuate fasciculus, and the optic radiations are in accordance with anatomical knowledge. The required user interaction is comparable to that of streamline tractography, which allows for an uncomplicated integration of the algorithm into the clinical routine.

ASSESSMENT OF THE LOG-EUCLIDEAN METRIC PERFORMANCE IN DIFFUSION TENSOR IMAGE SEGMENTATION

Introduction: Appropriate definition of the distance measure between diffusion tensors has a deep impact on Diffusion Tensor Image (DTI) segmentation results. The geodesic metric is the best distance measure since it yields high-quality segmentation results. However, the important problem with the geodesic metric is a high computational cost of the algorithms based on it. The main goal of this paper is to assess the possible substitution of the geodesic metric with the Log-Euclidean one to reduce the computational cost of a statistical surface evolution algorithm. Materials and Methods: We incorporated the Log-Euclidean metric in the statistical surface evolution algorithm framework. To achieve this goal, the statistics and gradients of diffusion tensor images were defined using the Log-Euclidean metric. Numerical implementation of the segmentation algorithm was performed in the MATLAB software using the finite difference techniques. Results: In the statistical surface evolution framework, the Log-Euclidean metric was able to discriminate the torus and helix patterns in synthesis datasets and rat spinal cords in biological phantom datasets from the background better than the Euclidean and J-divergence metrics. In addition, similar results were obtained with the geodesic metric. However, the main advantage of the Log-Euclidean metric over the geodesic metric was the dramatic reduction of computational cost of the segmentation algorithm, at least by 70 times. Discussion and Conclusion: The qualitative and quantitative results have shown that the Log-Euclidean metric is a good substitute for the geodesic metric when using a statistical surface evolution algorithm in DTIs segmentation.

Microstructural connectivity of the arcuate fasciculus in adolescents with high-functioning autism

The arcuate fasciculus is a white matter fiber bundle of great importance in language. In this study, diffusion tensor imaging (DTI) was used to infer white matter integrity in the arcuate fasciculi of a group of subjects with high-functioning autism and a control group matched for age, handedness, IQ, and head size. The arcuate fasciculus for each subject was automatically extracted from the imaging data using a new volumetric DTI segmentation algorithm. The results showed a significant increase in mean diffusivity (MD) in the autism group, due mostly to an increase in the radial diffusivity (RD). A test of the lateralization of DTI measurements showed that both MD and fractional anisotropy (FA) were less lateralized in the autism group. These results suggest that white matter microstructure in the arcuate fasciculus is affected in autism and that the language specialization apparent in the left arcuate of healthy subjects is not as evident in autism, which may be related to poorer language functioning.

A tracking‐based diffusion tensor imaging segmentation method for the detection of diffusion‐related changes of the cervical spinal cord with aging

To compare region of interest (ROI)-based and diffusion tensor tractography (DTT)-based methods for evaluating diffusion properties of the spinal cord as a function of age. Commonly, an ROI segmentation is used to delineate the spinal cord. In this work, new segmentation methods are developed based on DTT. In a first, DTT-based, segmentation approach, the diffusion properties are calculated on the tracts. In a second method, the diffusion properties are analyzed in the spinal cord voxels that contain a certain number of tracts. We studied the changes in diffusion properties of the human spinal cord in subjects of different ages. Diffusion tensor imaging (DTI) measurements of the cervical spinal cord were acquired on 42 healthy volunteers (age range = 19-87 years). The fractional anisotropy (FA), the mean diffusivity (MD), and eigenvalues (lambda(1), lambda(2), and lambda(3)) were compared for the ROI- and DTT-based segmentation methods. Our automatic techniques are shown to be highly reproducible and sensitive for detecting DTI changes. FA decreased (r = -0.38; P < 0.05), whereas MD and eigenvalues increased (r = +/- 0.45; P < 0.05) with age. These trends were not statistically significant for the ROI-based segmentation (P > 0.05). DTT is a robust and reproducible technique to segment the voxels of interest in the spinal cord.

A geometric flow-based approach for diffusion tensor image segmentation

Diffusion tensor magnetic resonance imaging (DT-MRI, shortened as DTI) produces, from a set of diffusion-weighted magnetic resonance images, tensor-valued images where each voxel is assigned a 3x3 symmetric, positive-definite matrix. This tensor is simply the covariance matrix of a local Gaussian process with zero mean, modelling the average motion of water molecules. We propose a three-dimensional geometric flow-based model to segment the main core of cerebral white matter fibre tracts from DTI. The segmentation is carried out with a front propagation algorithm. The front is a three-dimensional surface that evolves along its normal direction with speed that is proportional to a linear combination of two measures: a similarity measure and a consistency measure. The similarity measure computes the similarity of the diffusion tensors at a voxel and its neighbouring voxels along the normal to the front; the consistency measure is able to speed up the propagation at locations where the evolving front is more consistent with the diffusion tensor field, to remove noise effect to some extent, and thus to improve results. We validate the proposed model and compare it with some other methods using synthetic and human brain DTI data; experimental results indicate that the proposed model improves the accuracy and efficiency in segmentation.

DTI segmentation using an information theoretic tensor dissimilarity measure

In recent years, diffusion tensor imaging (DTI) has become a popular in vivo diagnostic imaging technique in Radiological sciences. In order for this imaging technique to be more effective, proper image analysis techniques suited for analyzing these high dimensional data need to be developed. In this paper, we present a novel definition of tensor "distance" grounded in concepts from information theory and incorporate it in the segmentation of DTI. In a DTI, the symmetric positive definite (SPD) diffusion tensor at each voxel can be interpreted as the covariance matrix of a local Gaussian distribution. Thus, a natural measure of dissimilarity between SPD tensors would be the Kullback-Leibler (KL) divergence or its relative. We propose the square root of the J-divergence (symmetrized KL) between two Gaussian distributions corresponding to the diffusion tensors being compared and this leads to a novel closed form expression for the "distance" as well as the mean value of a DTI. Unlike the traditional Frobenius norm-based tensor distance, our "distance" is affine invariant, a desirable property in segmentation and many other applications. We then incorporate this new tensor "distance" in a region based active contour model for DTI segmentation. Synthetic and real data experiments are shown to depict the performance of the proposed model.