AssemblyNet-AD-FTD produces a probability score based on deep learning models trained to discriminate Alzheimer's disease (AD) and Frontotemporal Dementia (FTD) signatures. The grading map provides a score for each structure that reflects the similarity of the subject under study to a population with AD or FTD (i.e., if the T1w MRI content presents AD-like or FTD-like patterns). In addition, AssemblyNet-AD-FTD combined deep grading based features with volume-based features to establish the final diagnosis.

All the considered structures are segmented using AssemblyNet

Be aware that AssemblyNet-AD-FTD has been designed only to discriminate AD and FTD patients from controls. Do not use it for other types of pathologies (e.g., Parkinson, Vascular dementia, ...). Moreover, the definition of AD follows ADNI and the definition of FTD follows the clinical classification proposed by NIFD. Only three sub-types of FTD are considered: behavioral variant FTD (bvFTD), semantic variant (SV) and progressive nonfluent aphasia (PNFA).