This article outlines imaging diagnosis of dementia, focusing on its usefulness in general clinical practice.

It also discusses caveats and limitations, including comparative information with pathological diagnosis.

For Alzheimer's disease (AD) and frontotemporal lobar degeneration (FTLD), we used the database, and the images were used from our own cases. For MRI-based diagnosis, we introduced voxel-based morphometry (VBM) and discussed the usefulness of machine learning. Although clinical diagnosis not always consistent with pathology, they were consistent in AD with a probability of more than 80%.

Machine learning based on the pattern of brain atrophy is expected to improve the accuracy of AD diagnosis. VBM may also be useful in the diagnosis of FTLD.