Application of neural network in dose calculation of radiotherapy

The precise conversion of CT numbers to their mass densities is essential in dose calculation of radiotherapy for the inhomogeneity. A new neural network algorithm was presented for tissue density calibration. Artificial neural networks are used to learn the relationships between the CT numbers and their mass densities. It neither requires an accurate mathematical model nor needs any priori knowledge about the parameters. This algorithm had been added to the convention DPM and can fulfill the requirement of heterogeneity correction in clinical radiotherapy. The high speediness and high precision are the features of this system. The results compared with PENELOPE proved the efficiency of this method.