Statistical Issues in the Analysis of Neuroimages

Abstract

Summary: Statistical issues in the analysis of neuroim ages are reviewed, These include biological questions of interest, basic problems of measurement and experimen tal design, normalisation, standardisation and transforma-

Mild and Moderate Hypothermia (α-Stat) Do Not Impair the Coupling Between Local Cerebral Blood Flow and Metabolism in Rats

“…Contrasts of slopes of the LCBF-LCGU regression lines were tested with common t test statistics with Bonferroni's correction for multiple comparisons. Because of the limitations of this kind of analysis, an additional, more rigorous statistical approach with log-transformed data was applied, in which the relationship of LCBF and LCGU was examined with a repeated measures ANOVA according to McCulloch et al 24 and Ford et al 25 For this analysis, a computer software package (BMDP2v, BMDP Statistical Software Inc) was used that provided a consideration of interanimal variability and enabled the detection of heterogeneities in the relationship between LCGU and LCBF.…”

“…Because observations are repeated on multiple regions of all animals and only the average values for each region are used, the real uncertainty about the relationship between glucose utilization and CBF might be greater than suggested by the regression analysis. 24 To exclude erroneous data interpretation with inappropriate statistical analysis, a second approach proposed by McCulloch et al 24 and Ford et al 25 was used, in which a repeated measures ANOVA was calculated with the use of log-transformed data.…”

Preclinical prediction of Alzheimer's disease using SPECT

“…55 The SVD approach was selected for the present analysis because of its potential strength in demonstrating results that are not derived using a priori knowledge of specific group differences (i.e., criterion values) or preselected morphologic boundaries. *,64 The relative strengths and weaknesses of these techniques have been discussed elsewhere.…”

Functional brain imaging of human sleep

“…At that time, the regional distribution of cerebral activity was not adequately described. This may be due to several factors: poor spatial resolution of the first generation scanners, low anatomical precision in the neuroanatomical placement of regions of interest (ROI), statistical limitations of the ROI method (Ford et al 1991), etc. All these reasons explain why, except for a thalamic deactivation in SWS (Maquet et al 1992;Maquet et al 1990) (Maquet et al 1992;Maquet et al 1990) while others described activation in other areas, including white matter regions (Buchsbaum et al 1989 reduced to 1 or 2 min, giving access to the study of more brief physiological processes.…”