Cluster analysis of diffusion tensor magnetic resonance images in human head injury

Citation
Dk. Jones et al., Cluster analysis of diffusion tensor magnetic resonance images in human head injury, NEUROSURGER, 47(2), 2000, pp. 306-313
Citations number
42
Categorie Soggetti
Neurology,"Neurosciences & Behavoir
Journal title
NEUROSURGERY
ISSN journal
0148396X → ACNP
Volume
47
Issue
2
Year of publication
2000
Pages
306 - 313
Database
ISI
SICI code
0148-396X(200008)47:2<306:CAODTM>2.0.ZU;2-5
Abstract
OBJECTIVE: issues surrounding the nature of the edema associated with traum atic brain injury in humans, and its evolution in the acute phase, remain u nresolved. This study aimed to characterize the topographical nature of the pathophysiological changes in human traumatic brain injury with diffusion tensor magnetic resonance imaging. METHODS: Multislice diffusion-weighted magnetic resonance imaging data were acquired from five patients undergoing elective ventilation for management of traumatic focal contusion or hematomas. The diffusion tensor and the Ta -weighted intensity were then computed for every voxel in the image data se t for each patient The topographical distribution of abnormalities in the t race of the diffusion tensor and T2-weighted images were characterized by c luster analysis. RESULTS: In four patients with technically satisfactory data, a narrow band of tissue was observed in the periphery of focal lesions, which was charac terized by selective reduction in the trace of the diffusion tensor, withou t any associated increase in the T2-weighted signal intensity. CONCLUSION: This change is interpreted as indicating either a partial redis tribution of water from the extra- to intracellular compartment, or a reduc tion in the diffusivity of water in the intracellular or cytosolic environm ent. These diffusion and T2-weighted characteristics are also found in earl y ischemic change, hence, such regions may represent potentially salvageabl e tissue at risk of permanent damage. The study illustrates the advantage o f using information contained within the diffusion tensor in addition to mo re conventional imaging sequences.