Abstract #2015

Assessment of mild cognitive impairment detection in a community-dwelling population using quantitative, multiparametric MRI-based classification

Mark J.R.J. Bouts^1,2,3, Jeroen van der Grond², Meike W. Vernooij^4,5, Tijn M. Schouten^1,2,3, Frank de Vos^1,2,3, Lotte G.M. Cremers^4,5, Mark de Rooij^1,3, Wiro J. Niessen^4,6,7, M. Arfan Ikram^4,5,8, and Serge A.R.B. Rombouts^1,2,3

¹Psychology, Leiden University, Leiden, Netherlands, ²Radiology, Leiden University Medical Center, Leiden, Netherlands, ³Leiden Institute for Brain and Cognition, Leiden University, Leiden, Netherlands, ⁴Radiology and Nuclear Medicine, Erasmus MC University Medical Center, Rotterdam, Netherlands, ⁵Epidemiology, Erasmus MC University Medical Center, Rotterdam, Netherlands, ⁶Medical Informatics, Erasmus MC University Medical Center, Rotterdam, Netherlands, ⁷Applied Sciences, Delft University of Technology, Delft, Netherlands, ⁸Neurology, Erasmus MC University Medical Center, Rotterdam, Netherlands

Multiparametric MRI-based classification algorithms improve classification of dementia over single measure classifications. Yet, how accurate these algorithms are in identifying subjects with mild cognitive impairment (MCI) in a general population is unclear. We evaluated single and multiparametric algorithms that include structural and diffusion tensor MRI in their potential to accurately differentiate MCI from normal aging subjects in a community-dwelling population. While highest classification rates were observed for multiparametric algorithms, overall classification performance was low (AUC: 0.524-0.631). Our results suggest that accurate MRI-based single subject detection of MCI within a population-based setting may be difficult to achieve using MR imaging alone.

This abstract and the presentation materials are available to members only; a login is required.

Join Here