Radiation hematologic toxicity prediction in rectal cancer: a comparative Radiomics-based study on CT image and dose map

Yingpeng Liu ^1* liping guo ¹ Yi Wang ¹ Qingtao Xu ² Jingfeng Zhang ¹ Xianyun Meng ¹

• ¹ Department of Gastroenterology, Ningbo No. 2 Hospital, Ningbo, China
• ² Ningbo First Hospital, Ningbo, Zhejiang Province, China

Background and Objectives: Acute radiation hematologic toxicity may disturb the radiotherapy plan and thus decrease the treatment outcome. However, whether the dose map has enough prediction value for detecting HT is still unknown. Methods: In the study, we collected the pre-treatment CT images and the in-treatment dose map from a discovery dataset of 299 patients and a validation dataset of 65 patients from another center. Then, the radiomic features of the clinical target volume (CTV) in the radiotherapy were respectively extracted, and the LASSO algorithm was used for feature dimension deduction and three classifiers including SVM (rbf kernel), random forest and Catboost were used respectively to construct the HT classification model in rectal cancer patients. The model performance was evaluated by both the internal 20% dataset and the external multicenter dataset.The results revealed that Catboost achieved the best model performance in almost all task, and CT images performed similarly with the dose map, although their combination model performed relatively lower. In addition, gender, age and some radiomic features from the decomposed image space were the most representative feature for HT prediction.Our study can confirm the HT occurrence in LARC patients was multifactorial, and combining effective features together can classify the high-risk patients with HT, thus timely preventing or intervening HT to improve the subsequent outcome.