According to who data , Alzheimer's disease （AD） Is the most common cause of dementia , Of dementia patients 70%. All over the world , There are about 2400 Ten thousand people were affected , And this figure is expected every 20 It will double in a year .
Researchers at Kaunas University in Lithuania have developed a method based on deep learning , Sure Predicting the onset of Alzheimer's disease from brain images , Accuracy over 99%.
“ Medical experts around the world are trying to raise awareness of the early diagnosis of Alzheimer's disease , This provides patients with a better opportunity to benefit from treatment . This is a doctoral student from Nigeria Modupe Odusami（ The first author of the study ） One of the most important reasons for choosing this topic .”Odusami My doctoral supervisor 、 Kaunas University of Technology （KTU） Researcher of the Department of Multimedia Engineering, School of information Rytis Maskeliūnas say .
This model based on deep learning was developed by Lithuanian researchers in the field of artificial intelligence , They changed it 18 Layer depth residual network （ResNet18）, Yes 138 Of subjects fMRI Image classification .
These images are Divided into six different categories ： From health to MCI To the spectrum of Alzheimer's disease .
In total, researchers from the Alzheimer's neuroimaging program fMRI Data set I chose 51443 Zhang he 27310 Images are trained and verified .
The model can be used in a given data set Find effectively MCI Characteristics of .
For the early MCI And AD、 Advanced MCI And AD、MCI And early MCI, Respectively reached 99.99%、99.95% and 99.95% The best classification accuracy .
Address of thesis ：https://www.mdpi.com/2075-4418/11/6/1071