Radiobiological modeling of pituitary gland complications following head and neck radiotherapy using conventional models and artificial neural network

Abstract

In this study, using radiobiological modeling and artificial neural network, the normal tissue complication probability (NTCP) in radiotherapy of nasopharynx and brain tumors after three-dimensional conformal treatment was calculated. Materials and Methods: 20 patients with nasopharynx and 31 patients with brain tumors treated with 3D-conformal method were studied. Necessary data including minimum, maximum and mean dose as well as number of treatment sessions and prescribed dose were extracted from the dose-volume histogram for each patient from the treatment planning system.Two radiobiological models, LKB and Log-logistic, were used to calculate the NTCP. Also, an artificial neural network was used to calculate the NTCP of patients. Results: The mean dose of pituitary gland for patients with nasopharynx and brain tumor was 30.42 and 51.29 Gy respectively.The mean NTCP calculated by LKB and Log-logistic models for nasopharyngeal patients was 54.53% and 50.83%, respectively. These values are 62.23% and 53.55% for brain tumor patients, respectively. In the artificial neural network, the error rate was 0.000383 for the training part and 2.21E-5 for the test part. The mean squared (R2) was 0.98 for the training part and 0.99 for the test part.