• Bing Zhang, Ming Li, Zong-Zheng Sun, Bin Zhu, Lei Yuan, Yi Wang, and Yun Xu.
• Department of Radiology, The Affiliated Drum Tower Hospital of Nanjing University Medical School, Nanjing, China.
• J Clin Neurosci. 2009 May 1; 16 (5): 635-41.

AbstractWe aimed to investigate the use of advanced functional MRI (fMRI) techniques such as proton magnetic resonance spectroscopy ((1)H-MRS) and the apparent diffusion coefficient (ADC) value in diffusion weighted imaging (DWI), in the diagnosis of mild cognitive impairment (MCI). Multiple indicators were combined in order to improve the early diagnostic value of MRS and ADC. We administered MRS and DWI-ADC to 13 patients with Alzheimer's disease (AD), 9 patients with MCI, and 13 control patients. Changes in N-acetylaspartate/creatine and phosphocreatine (NAA/Cr), myoinositol/creatine (mI/Cr), and the ADC values in the hippocampus and the temporoparietal region were compared among groups. The sensitivity and specificity of different markers were analyzed individually and combined with others. All participants were evaluated by the mini mental state examination (MMSE), and the correlation between NAA/Cr, MI/Cr, ADC and the score of MMSE were analyzed separately. The NAA/Cr, mI/Cr and ADC values in the hippocampus among AD, MCI patients, and controls were significantly different (p<0.05). At a fixed specificity of 84.6%, the high sensitivity of 100% and 92.9% in differentiating AD and MCI from normal controls were obtained by combining the three indicators. The receiver operating characteristic plots illustrated that the area under the multimarker curve was the biggest among the all four curves, and the sensitivity of the multimarkers was highest. The best correlation was obtained between ADC and MMSE, rather than between NAA or mI and MMSE. Thus, we found that changes in NAA/Cr, mI/Cr and ADC in the hippocampus and the temporoparietal regions were helpful in the clinical diagnosis of MCI. Furthermore, these changes showed potential in predicting the progression of MCI to AD if the multimarkers were combined.

30 keywords...

hide…