To investigate the feasibility of assessing the treatment response using diagnostic-quality CT imaging features during radiotherapy for esophageal cancer.

Thirty-three patients with stage Ⅰ to IV esophageal cancer undergoing intensity-modulated radiotherapy were recruited in this study. CT images were acquired using a CT-on-rail imaging system. Imaging data of CT images including gross tumor volume (GTV), the volume of spinal cord and non-irradiated tissue (NIT), CT mean (MCTN), standard deviation , and skewness were collected and analyzed by using MIM image processing system. Patients were divided into the effective group (complete remission and partial remission, n=24) and ineffective group (no change and progression, n=9) based on the outcomes of 3-month follow-up. The imaging data were statistically compared between two groups using the self-designed Matlab software.

The tumor volume and MCTN of 33 patients were gradually decreased with the increase of radiotherapy dose. The tumor volume and MCTN were decreased by 42.46% and 5.76 HU in the effective group, more significant compared with 21.76% and 3.66 HU in the ineffective group (both P<0.005). The skewness in the ineffective group was decreased by 0.503 with the increasing radiation dose, whereas that in the effective group was increased by -0.450(P=0.034). Spinal cord and NIT did not significantly change with the increasing radiation dose.

Analysis of the characteristic data of CT images of patients with esophageal cancer during radiotherapy may early predict the clinical efficacy of radiotherapy.

Radiomics; Early assessment; Esophageal neoplasm/radiotherapy