Conventional magnetic resonance imaging (MRI) of brain and spinal cord is an essential diagnostic tool to exclude amyotrophic lateral sclerosis (ALS) "mimics", when a plausible alternative neuroanatomical localization is suspected 1 . The finding of hyperintensities along the course of corticospinal tracts (CST) on T2, Proton density (PD) or fluid-attenuated inversion recovery (FLAIR) was documented over 25 years ago in case reports and subsequent small series. The sensitivity of such changes has commonly been estimated to be less than 40% and the specificity is also poor, typically less than 70% 2-12 . Poor specificity values reflect the occurrence of similar hyperintensities in normal controls 13,14 , as well as widely varying conditions, such as leukodystrophies 15,16 , X-linked CharcotMarie-Tooth disease 17 , end stage cirrhosis and liver transplantation 18 . A 2010 guideline from the European Federation of Neurological Societies states that such hyperintensities may help to support a suspicion of ALS but that ABSTRACT: Background: There is currently no definite neuroimaging test to detect amyotrophic lateral sclerosis (ALS), which leads to significant delay in diagnosis, particularly if one takes into account the rapidity of disease evolution. Hyperintensity of the corticospinal tracts (CST) on T2 or fluid-attenuated inversion recovery (FLAIR) weighted magnetic resonance imaging (MRI) has been well described, but data on sensitivity and specificity in larger series is lacking to help guide its application to clinical care. Methods:We analyzed clinical and MRI data from 64 patients with a definite retrospective diagnosis of ALS. In this case-control study, two experienced blinded neuroradiologists systematically assessed defined rostrocaudal segments of the intracranial course of the CST. Results: The overall sensitivity and specificity of conventional MRI for the diagnosis of ALS were 48% and 76% respectively. Highest specificities for CST hyperintensity were noted for the subcortical white matter (92%), centrum semiovale (88%) and medullary pyramids (92%). The lowest specificities were found for the cerebral peduncle (36%) and internal capsule (32%). We did not find a correlation with the rate of clinical progression, age of onset or the presence of upper motor neuron signs on examination. Conclusion: Conventional MRI was not found to be a reliable diagnostic tool for ALS and it did not help predict clinical characteristics such as speed of evolution or prominence of upper motor neuron signs. Its main role in the setting of ALS should remain to help exclude alternative diagnostic considerations. A multimodal approach relying on newer functional and structural MRI techniques still needs to be developed and validated. RÉSUMÉ: Précision de l'IRM conventionnelle dans la SLA. Contexte : Il n'existe pas actuellement de test de neuroimagerie pour détecter la SLA, ce qui occasionne des délais importants dans le diagnostic de la maladie, particulièrement si on tient compte de la rapidité d'évolution de celle-ci. Un...