Impact Factor 4.137 | CiteScore 4.28
More on impact ›

Original Research ARTICLE Provisionally accepted The full-text will be published soon. Notify me

Front. Oncol. | doi: 10.3389/fonc.2019.01250

Radiomics analysis of iodine-based material decomposition images with dual energy CT imaging for preoperatively predicting microsatellite instability status in colorectal cancer

Jingjun Wu1, Qinhe Zhang1, Ying Zhao1, Yijun Liu1, Anliang Chen1, Xin Li2, Tingfan Wu2, Jianying Li3,  Yan Guo3 and  Ailian Liu1*
  • 1Department of Radiology, First Affiliated Hospital, Dalian Medical University, China
  • 2Translational Medicine Team, GE Healthcare (China), China
  • 3GE Healthcare (China), China

Purpose: To investigate the value of radiomics analysis of iodine-based material decomposition (MD) images with dual energy computed tomography (DECT) imaging for preoperatively predicting microsatellite instability (MSI) status in colorectal cancer (CRC).
Methods: This study included 102 CRC patients proved by postoperative pathology, and their MSI status was confirmed by immunohistochemistry staining. All patients underwent preoperative DECT imaging scanned on either a Revolution CT or Discovery CT 750HD scanner, and the iodine-based MD images in the venous phase were reconstructed. The clinical, CT-reported and radiomics features were obtained and analyzed. Data from the Revolution CT scanner were used to establish a radiomics model to predict MSI status (70% samples were randomly selected as the training set, and the remaining samples were used to validate); and data from the Discovery CT 750HD scanner were used to test the radiomics model. The stable radiomics features with both inter-user and intra-user stability were selected for next analysis. The feature dimension reduction was performed by using Student’s t test or Mann-Whitney U test, Spearman’s rank correlation test, Min-Max standardization, one-hot encoding, and least absolute shrinkage and selection operator selection method. The multi-parameter logistic regression model was established to predict MSI status. The model performances were evaluated: the discrimination performance was accessed by receiver operating characteristic curve (ROC) analysis; the calibration performance was tested by calibration curve accompanied by Hosmer-Lemeshow test; the clinical utility was assessed by decision curve analysis.
Results: Nine top-ranked features were finally selected to construct the radiomics model. In the training set, the area under ROC curve (AUC) was 0.961 (accuracy: 0.875; sensitivity: 1.000; specificity: 0.812); in the validation set, the AUC was 0.918 (accuracy: 0.875; sensitivity: 0.875; specificity: 0.857). In the testing set, the diagnostic performance was slightly lower with AUC of 0.875 (accuracy: 0.788; sensitivity: 0.909; specificity: 0.727). Nomogram based on clinical factors and radiomics score was generated via the proposed logistic regression model. Good calibration and clinical utility were observed using the calibration and decision curve analysis, respectively.
Conclusion: Radiomics analysis of iodine-based MD images with DECT imaging holds great potential to predict MSI status in CRC patients.

Keywords: Microsatellite Instability, Colorectal Neoplasms, Iodine-based material 2 decomposition image, Radiomics, Dual energy computed tomography

Received: 29 Jul 2019; Accepted: 30 Oct 2019.

Copyright: © 2019 Wu, Zhang, Zhao, Liu, Chen, Li, Wu, Li, Guo and Liu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

* Correspondence: Prof. Ailian Liu, Department of Radiology, First Affiliated Hospital, Dalian Medical University, Dalian, Liaoning Province, China, liuailian@dmu.edu.cn