Estimation of the glucose metabolism from dynamic PET-scans using neural networks

A method for fast pixel by pixel estimation of the glucose metabolism in the brain using the tracer [/sup 18/F]fluorodeoxy-glucose in dynamic positron emission tomography (PET)-scan data is described. A neural network is trained to estimate the glucose metabolism on data generated by direct fitting of the rate constants in Sokoloff's model. The generalisation ability of the neural network is tested on data from subjects not included in the training set. This method can be used to estimate changes of the metabolism in different brain regions for subjects with serious brain disorders. By using the neural estimation procedure the processing time for a brain scan volume is reduced from 48 hours to 4 minutes.