作者: W. Haenicke , A. Kleinschmidt , J. Frahm , K. D. Merboldt , M. Requardt
DOI:
关键词: Medicine 、 Robustness (computer science) 、 Artificial intelligence 、 Dynamic contrast-enhanced MRI 、 Temporal resolution 、 Functional magnetic resonance spectroscopy of the brain 、 Sensitivity (control systems) 、 Human brain 、 Pattern recognition 、 Magnetic resonance imaging 、 Subtraction
摘要: The sensitivity of gradient-echo magnetic resonance imaging (MRI) to changes in cerebral blood oxygenation has been introduced for mapping functional brain activation. To benefit from the high spatial and temporal resolution respective dynamic MRI data sets, their analysis requires algorithms that are capable both precisely delineating task-related activation patterns demonstrating connectivity interacting areas. Here, we present various strategies evaluation by means correlational analyses surpass quality subtraction-based maps improving robustness. On a pixel-by-pixel basis approach correlates signal time courses with reference function, reflecting sequence activated control states. Extended versions employ calculation auto- or cross-correlation functions increase sensitivity, but require periodic stimulations. Following individual correction non-specific correlated fluctuations, coherent can be improved using neighborhood principles. Such refined expected enhance usefulness oxygenation-sensitive studying anatomy human under physiological pathological conditions.