Prevalence of vascular dementia in Brazil and Latin America is higher than in the global north (e.g. US and Europe) probably due to socio-economical factors [1, 2]. Dementia represents a huge economical challenge for Brazil and South America given that in these countries the demographical pyramid is starting to invert. In contrast to other types of dementia, vascular dementia, which mostly affects the lower-income population, can be prevented. Nevertheless research in vascular dementia is not very extensive.
In São Paulo we have access to big MRI data resources from the public health system, given a relatively high density of MRI scanners and of population in comparison to other areas of Brazil and other Latin American countries. This data, which represents mostly the lower income target group, can be accessed. It is important to highlight that the distribution of MRI scanners in Brazil is very heterogeneous. While some states have the same ratio of MRI scanners per 100.000 habitants as in some European countries, other states of Brazil still do not reach the minimum recommended by the WHO.
The challenge is to generate a system that can store clinical data known as risk factors for vascular dementia such as for example blood pressure, glycemia, cholesterol together with MRI data (e.g. DICOM). Ideally MRI images should be processed automatically in order to quantify: T2 hyperintensities, ischemic lesions, and hemorrhagic lesions (T2* or SWI). Machine learning algorithms could then be applied to the data in order to establish a gender and age specific threshold for clinical data at which significant brain changes (correlated with a brain MR scan) start to occur. The final goal would be that, based on this data, action guidelines could be created for patients without access to MRI, but with other clinical data exceeding the threshold. These patients could then be referred to a brain MRI that would allow an early intervention to prevent the development of vascular dementia.
Figure 1: An example of MRI of a patient with vascular dementia.
What would be the best way technically to generate such a system (software, data format, algorithms, MR parameters)?
References:  Grinberg LT, Nitrini R, Suemoto CK, et al. Prevalence of dementia subtypes in a developing country: a clinicopathological study. Clinics. 2013;68(8):1140-1145.  Kalaria, R. N., Maestre, G. E., Arizaga, et al. Alzheimer’s disease and vascular dementia in developing countries: prevalence, management, and risk factors. The Lancet Neurology 2008, 7(9), 812-826.