Quantitative analysis of functional images requires a strategy for reducing to tractable size the overwhelming amount of numeric data contained therein. Region-of-interest (ROI) selection is the most widely used means of image-data reduction, but it has many limitations. Spatial bias, introduced by selection of regions as being ''of interest,'' is probably the greatest limitation of ROI analyses. Change-distribution analysis is a new data-analysis strategy that eliminates this a priori selection bias in a way that can increase the sensitivity, specificity, and localization precision. All image pixels are surveyed for changes from the control condition. Only areas of change are sampled and contribute to statistical analysis. Change-distribution analysis has been validated for within-subject pairs of images, but it is potentially applicable in a wide variety of imaging protocols.
- Change-distribution analysis
- Positron emission tomography (PET)
- Region of interest (ROI)
ASJC Scopus subject areas
- Clinical Neurology
- Cardiology and Cardiovascular Medicine