ArcheD, a residual neural network for prediction of cerebrospinal fluid amyloid-beta from amyloid PET images.

A convolutional neural network was developed to predict Aβ CSF from Aβ PET scans. Predicted Aβ CSF values significantly correlated with cortical Aβ standardized uptake value ratio and episodic memory.White matter was the most relevant region for the model's prediction of Aβ CSF.Cerebral white matter contributed more to prediction of amyloid CSF on early AD stages. Gray matter contributed more to prediction on late AD stages, especially temporal lobe.