Original Investigation High Order Diffusion Tensor Imaging in Human Glioblastoma Benjamin M. Ellingson, PhD, Timothy F. Cloughesy, MD, Albert Lai, MD, PhD, Phioanh L. Nghiemphu, MD, Linda M. Liau, MD, Whitney B. Pope, MD, PhD Rationale and Objectives: Diffusion tensor imaging has been used to characterize tumor heterogeneity and invasion in human glioblas- toma. Recently, higher order diffusion tensors have been proposed as solutions to errors associated with diffusion tensor imaging esti- mates of complex microstructures. The purpose of the current study was to examine higher order diffusion characteristics in human glioblastoma prior to surgical resection using the fourth-order diffusion tensor model. Materials and Methods: Twenty-five patients with newly diagnosed glioblastoma participated in the study. Diffusion-weighted images were collected in 21 directions. The second-order (traditional) and fourth-order diffusion tensors were calculated and compared in regions of contrast enhancement, T2 signal abnormality, and normal-appearing white matter. Results: Orientation distribution functions were strikingly different between the two tensor models, particularly in regions with tumor heterogeneity and/or regions of suspected tumor invasion. Image contrast was significantly higher in fourth-order scalar measures compared to second-order scalars. Results of particular eigenvalues and scalars using the fourth-order tensor showed differences between T2 abnormal regions and contrast enhancement, whereas second-order eigenvalues and scalars did not show differences. This suggests that higher order diffusion images could potentially be more sensitive to tumor invasion. Conclusions: These results suggest that the fourth-order diffusion tensor has the ability to add value to second-order (traditional) diffusion tensor imaging in the evaluation of glioblastoma. Key Words: DTI; diffusion MRI; fourth-order tensor; GBM; glioblastoma; brain tumor. ªAUR, 2011 D iffusion-weighted magnetic resonance imaging (MRI) techniques are highly sensitive to the under- lying microstructural characteristics of biologic tissues. This sensitivity to subvoxel, microscopic features has helped provide insight into many physiologic changes that occur as a result of brain tumor growth and invasion, such as cerebral edema (1), hypoxia (2), the increase in diffusion observed after successful radiotherapy due to cell breakdown (3), and the change in diffusion characteristics resulting from increasing tumor cellularity (4) and invasion (5,6). Additionally, diffusion magnetic resonance characteristics have been shown to be predictive (7,8) and prognostic (6,9) biomarkers in new brain tumor therapeutics and have shown utility in histopathologic grading of gliomas (10). Diffusion tensor imaging (DTI) involves the addition of directional encoding to diffusion measurements, allowing novel structural information about the microenvironment to be acquired. For example, in normal tissues, DTI typically shows high diffusion anisotropy within tightly packed white matter fiber bundles because of diffusion restriction perpen- dicular to axon fibers. This high degree of diffusion anisotropy within white matter regions provides the basis for DTI tractog- raphy (11), in which pseudoaxonal tracts are ‘‘grown’’ from seed regions placed within white matter tracts. For relatively simple tissue structures, such as the thick white matter bundle within the corpus callosum, the ‘‘traditional’’ diffusion tensor model may be an adequate representation of the general tissue architecture. For more complex tissues, ‘‘nontraditional’’diffu- sion models may be beneficial. Primary human brain tumors, such as the highly aggressive and malignant glioblastoma, have an extremely complex and heterogeneous microenvironment consisting of pallisading necrosis, edema, leaky neovasculature, and cells of various sizes excreting numerous signaling molecules and proteins. Traditional DTI techniques have shown tremendous utility in the diagnosis (12,13), prognosis (14), and surgical planning of adult primary brain tumors (15,16). Traditional DTI involves collecting multiple diffusion-weighted images, encoded for specific directional sensitivities, and then fitting these data to a 3  3, second-order diffusion tensor field (17). Higher order diffusion tensors, such as the fourth- Acad Radiol 2011; -:1–8 From the Department of Radiological Sciences (B.M.E., W.B.P.), the Department of Neurology (T.F.C., A.L., P.L.N.), and the Department of Neurosurgery (L.M.L.), David Geffen School of Medicine, University of California, Los Angeles, 10833 Le Conte Ave, BL-428 CHS, Los Angeles, CA 90095-1721 Received January 28, 2011; accepted February 25, 2011. Dr Pope was supported by the Brain Tumor Funders Collaborative. Dr Cloughesy was supported by Art of the Brain, the Ziering Family Foundation in memory of Sigi Ziering, the Singleton Family Foundation, and the Clarence Klein Fund for Neuro-Oncology. Address correspondence to: B.M.E. e-mail: bellingson@mednet.ucla.edu ªAUR, 2011 doi:10.1016/j.acra.2011.02.020 1 FLA 5.1.0 DTD  XACRA2503_proof  21 April 2011  2:38 pm  ce 68 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113