Premium
P1‐365: DIFFERENTIATING TAUOPATHIES WITH A NOVEL DIFFUSION TENSOR IMAGING METHOD TO INVESTIGATE CORTICAL ARCHITECTURE
Author(s) -
Chance Steven A.,
Torso Mario,
Jenkinson Mark
Publication year - 2019
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2019.06.920
Subject(s) - frontotemporal dementia , progressive supranuclear palsy , diffusion mri , receiver operating characteristic , primary progressive aphasia , semantic dementia , frontotemporal lobar degeneration , tauopathy , fractional anisotropy , psychology , neuroscience , audiology , dementia , medicine , pathology , neurodegeneration , disease , radiology , magnetic resonance imaging
classifications were performed twice (within 12 weeks): first with standard radiological reading and then using SVM weight maps as a guide. Balanced accuracies were computed as classification metrics. Generalized Linear Mixed Model (GLMM) with logit link were performed for each diagnostic pairs to investigate the potential improvement of each radiologist using the SVM weight maps. Results: Diagnostic accuracies are presented in Table 2. Diagnostic performance was statistically improved by the use of the weight maps for the two junior unspecialized radiologists in the case of FTD vs EOAD. Conclusions: We demonstrate that using outputs from an artificial intelligence tool can improve the diagnostic accuracy of junior radiologists. Notably, the improvement is over 10 points of accuracy for FTD vs EOAD, a clinically relevant and difficult diagnosis. SVMweight maps are easy to integrate in the clinical routine workflow and could thus contribute to improve the diagnosis of dementia.