Early Prediction of Xerostomia During Radiation Therapy for Head and Neck Cancer Based on Radiomic Analysis of Daily CT

Abstract

To investigate changes of CT radiomicss of parotid glands (PG) to predict acute Xerostomia during radiotherapy (RT) for head and neck cancer (HNC). Daily diagnostic-quality CTs were acquired for 23 HNC patients. PG contours were generated and changes of histogram-based radiomics were calculated. Correlations between metrics, radiation dose, and observed acute Xerostomia were analyzed. A CT-Based Xerostomia Score (CTXS) is introduced to evaluate/predict xerostomia by combining changes of mean CT number (MCTN) and volume of parotid glands. Substantial changes in various radiomic metrics of PGs were observed. The average MCTN and volume for all patients were reduced by 9.4 HU and 49.6 cm3, respectively. Individually, the changes of PG MCTN or volume are not strongly correlated with the observed xerostomia grades while the CTXS did. The xerostomia grade at the end of treatment can be predicted based on the CTXS determined at the 5th week with a precision of 80% and a sensitivity of 100%. Significant changes in the CT histogram radiomic metrics of the parotid glands were observed during RT of HNC. A practical method of using the changes of MCTN and volume is proposed to predict radiation-induced acute xerostomia, allowing adaptive treatment.