Novel region of interest interrogation technique for diffusion tensor imaging analysis in the canine brain. Issue 4 (August 2017)

• Record Type: Journal Article
• Authors: Li, Jonathan Y; Middleton, Dana M; Chen, Steven; White, Leonard; Ellinwood, N Matthew; Dickson, Patricia; Vite, Charles; Bradbury, Allison; Provenzale, James M
• Is Part Of: Neuroradiology journal. Volume 30:Issue 4(2017:Aug.)
• Page Start: 339
• Page End: 346
• Publication Date: 2017-08
• Subjects: Diffusion tensor imaging -- magnetic resonance -- canine model -- region of interest -- interrogation technique -- reproducibility; Nervous system -- Radiography -- Periodicals; Neuroradiography -- Periodicals; Electronic journals; 616.804757
• DOI: 10.1177/1971400917709629
• Languages: English
• ISSNs: 1971-4009

Abstract

Purpose: We describe a novel technique for measuring diffusion tensor imaging metrics in the canine brain. We hypothesized that a standard method for region of interest placement could be developed that is highly reproducible, with less than 10% difference in measurements between raters. Methods: Two sets of canine brains (three seven-week-old full-brains and two 17-week-old single hemispheres) were scanned ex-vivo on a 7T small-animal magnetic resonance imaging system. Strict region of interest placement criteria were developed and then used by two raters to independently measure diffusion tensor imaging metrics within four different white-matter regions within each specimen. Average values of fractional anisotropy, radial diffusivity, and the three eigenvalues (λ1, λ2, and λ3) within each region in each specimen overall and within each individual image slice were compared between raters by calculating the percentage difference between raters for each metric. Results: The mean percentage difference between raters for all diffusion tensor imaging metrics when pooled by each region and specimen was 1.44% (range: 0.01–5.17%). The mean percentage difference between raters for all diffusion tensor imaging metrics when compared by individual image slice was 2.23% (range: 0.75–4.58%) per hemisphere. Conclusion: Our results indicate that the technique described is highly reproducible, even when applied to canine specimens of differing age, morphology, and image resolution. We propose this technique for future studies of diffusion tensor imaging analysis in canine brains and for cross-sectional and longitudinal studies of canine brain models of human central nervous system disease.