The impact of the novel CovBat harmonization method on enhancing radiomics feature stability and machine learning model performance: A multi-center, multi-device study.

Journal: European journal of radiology
PMID:

Abstract

PURPOSE: This study aims to assess whether the novel CovBat harmonization method can further reduce radiomics feature variability from different imaging devices in multi-center studies and improve machine learning model performance compared to the ComBat method.

Authors

  • Chuanghui Zhou
    Department of Imaging Diagnosis, Nanfang Hospital, Southern Medical University, Guangzhou 510000, Guangdong, China; School of Medical and Information Engineering, Gannan Medical University, Ganzhou 341000, Jiangxi, China.
  • Jianwei Zhou
    Institute of Physical Education, Xuchang University, Xuchang, Henan 461000, China.
  • Yijun Lv
    School of Medical and Information Engineering, Gannan Medical University, Ganzhou 341000, Jiangxi, China.
  • Maidina Batuer
    School of Biomedical Engineering, Southern Medical University, Guangzhou 510000, Guangdong, China.
  • Jinghan Huang
    College of Economics and Management, Shanghai Ocean University, Shanghai 201306, China.
  • Junyuan Zhong
    Medical Imaging Department of Ganzhou People's Hospital, Ganzhou 341000, Jiangxi, China.
  • Haijian Zhong
    School of Medical and Information Engineering, Gannan Medical University, Ganzhou 341000, Jiangxi, China. Electronic address: hjzhong2007@gmu.edu.cn.
  • Genggeng Qin
    Department of Radiology, Nanfang Hospital, Southern Medical University, Guangzhou, China.