Apport de deux techniques d’analyse automatisées des examens [ 18F] FDG TEP dans la discrimination des patients déments frontotemporaux et malades d’Alzheimer

Abstract

Introduction In younger patients, the in vivo clinical diagnosis of Alzheimer's disease (AD) and of the frontotemporal type (FTD) may be cumbersome. The gold standard diagnostic proof is currently still based upon pathology examination. It is crucial to find reliable techniques to make an accurate in vivo diagnosis and to differentiate the etiology of the dementia. Patients and method Twenty-four patients bearing clinically diagnosed AD ( n = 16) and FTD ( n = 8) underwent [ 18F] FDG-PET/CT brain scan. Four nuclear medicine physicians with varying expertise in neuroimaging read each scan according to: visual analysis; automated analysis computed by BRASS ® Hermes ® software; automated analysis computed by Cortex ID ® General Electric ® software. Interpretation aimed at assessing the global scan aspect, the cerebral metabolism per hemisphere (in five relevant regions) and the diagnostic degree of confidence. Diagnostic interpretations derived from visual and automated analyses were compared to clinical diagnosis. Inter-observer agreement and Kappa scores were calculated. Results Kappa analyses showed a gain in diagnostic accuracy for a nonexpert physician, a gain in diagnostic confidence with Cortex ID ® and a gain in interobserver diagnostic agreement with BRASS ®. Conclusion Using automated software such as Cortex ID ® or BRASS ® helps standardizing the interpretation of [ 18F] FDG distribution pattern in AD or FTD.