Program Information
Prediction of Acute Xerostomia Based On CT Texture Changes During Radiation Therapy for Nasopharyngeal Cancer
X Yang1,2,a)*, X Chen1,a), H WU3,a), Y TAO1,2,a) , H CHANG2 , X Deng2 , Y XIA2,b) , X Li1,b), (1) Sun Yat-Sen University Cancer Center (SYSUCC), Guangzhou City, Guangdong Prov., (2) Medical College of Wisconsin, Milwaukee, WI, (3) The Affiliated Cancer Hospital of Zhengzhou University, Zhengzhou, Henan. a) The authors have the equal contribution as the first author; b) Corresponding authors
Presentations
TH-AB-201-8 (Thursday, August 3, 2017) 7:30 AM - 9:30 AM Room: 201
Purpose: A model, CT-based Xerostomia Score (CTXS), for predicting acute xerostomia based on radiomics features of daily-CT acquired during radiotherapy for head and neck cancer was previously proposed. This study aims to validate the model using an independent dataset of nasopharyngeal cancer (NPC) patients.
Methods: CT and outcome data for 15 NPC patients treated with concurrent chemo-radiotherapy (12/15) or IMRT alone (3/15) with a prescription dose of 68.1 Gy in 30 fractions were analyzed. The patients ages ranged 24–72y with tumor stages of I-IV. For each patient, five CT sets were acquired at the treatment position at 0th, 10th, 20th, 30th fractions during RT, and at 3-month after RT. The parotid glands (PG) were delineated on each CT set by a radiation oncologist and verified independently by another one. Acute xerostomia was evaluated based on RTOG acute toxicity scoring, and six radiomics features were calculated for PG from each CT set, including the mean CT number (MCTN) in Hounsfield Unit (HU), standard deviation, volume, skewness, kurtosis, and entropy. The CTXS obtained for each patient were used to evaluate and predict the acute xerostomia level at 10th and 30th fractions.
Results: Substantial changes in various radiomics metrics of PG during radiotherapy were observed. The average MCTN and volume for all 15 patients were reduced by -11.5±13.7HU and -24.4±11.0% (10th fractions); -8.9±19.0HU and -47.8±13.7% (30th fractions), respectively. The CTXS model predicts the level of acute xerotomia with a precision of 96.7% (29/30) and a sensitivity of 100%, as compared to the clinical observed level.
Conclusion: This study demonstrates that radiation-induced acute xerotomia level can be predicted based on the observed changes in MCTN and volume of the PG during IMRT delivery. The CTXS model which is validated with two independent datasets could be used to robustly predict acute xerotomia for individual patient.
Contact Email: