همایش بین المللی هوش مصنوعی

صفحه اصلی / اولین همایش بین المللی هوش مصنوعی

Brain Age Classification from fMRI Data Using Graph Neural Networks and Evolutionary Algorithm

نویسندگان :

Nastaran Hassanzadeh¹ Mohammad Saniee Abadeh²

1- Faculty of Electrical and Computer Engineering Tarbiat Modares University 2- Faculty of Electrical and Computer Engineering Tarbiat Modares University

کلمات کلیدی :

brain-age،GNN،EA،fMRI

چکیده :

The brain is a complex organ that undergoes changes with age and prediction of brain age is very important for monitoring brain health, as it can provide useful information about the brain and help people to prevent neurological diseases. This research predicts brain age through age classification based on fMRI data from the HCP dataset of individuals between 22 and 36 years old. After the training phase on the graph convolutional neural network, the model’s accuracy on the test data reached 0.73, showing an improvement over previous works on the same dataset. Then, an evolutionary approach was used to optimize the selection of brain regions. For this purpose, a Genetic Algorithm was applied to discover important and knowledgeable regions. This selection and optimization process maintains good predictive accuracy while reducing the number of brain regions. The results show that despite using half the original number of brain regions, 8 regions, the accuracy of model is 0.65, which shows only a slight degradation, while it indicates the importance of these regions for brain age classification. Identifying these key regions can be effective in the early diagnosis of brain and neurological diseases, allowing experts to better understand and manage the brain aging process.