Z Kong1, J Li2, Zehua Liu3, Zhenyu Liu4, D Zhao5, X Cheng6, L Li3, Y Lin7, Y Wang8, J Tian9, W Ma10. 1. Department of Neurosurgery, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China; Department of Nuclear Medicine, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China. 2. Department of Neurosurgery, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China; School of Medicine, Tsinghua University, Beijing, China. 3. Collaborative Innovation Center for Internet Healthcare, Zhengzhou University, Zhengzhou, Henan, China. 4. CAS Key Laboratory of Molecular Imaging, Institute of Automation, Chinese Academy of Science, Beijing, China; University of Chinese Academy of Sciences, Beijing, China. 5. Department of Pathology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China. 6. Department of Nuclear Medicine, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China. 7. Collaborative Innovation Center for Internet Healthcare, Zhengzhou University, Zhengzhou, Henan, China; School of Software, Zhengzhou University, Zhengzhou, Henan, China. 8. Department of Neurosurgery, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China. Electronic address: ywang@pumch.cn. 9. CAS Key Laboratory of Molecular Imaging, Institute of Automation, Chinese Academy of Science, Beijing, China; University of Chinese Academy of Sciences, Beijing, China; Beijing Advanced Innovation Center for Big Data-Based Precision Medicine, School of Medicine, Beihang University, Beijing, China; Engineering Research Center of Molecular and Neuro Imaging of Ministry of Education, School of Life Science and Technology, Xidian University, Xi'an, Shaanxi, China. Electronic address: jie.tian@ia.ac.cn. 10. Department of Neurosurgery, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China. Electronic address: mawb2001@hotmail.com.
Abstract
AIM: To investigate a radiomics method based on 2-[18F]-fluoro-2-deoxy-d-glucose (FDG) positron-emission tomography (PET) to non-invasively evaluate proliferative activity in gliomas. MATERIALS AND METHODS: A total of 123 patients with histopathologically confirmed primary glioma were reviewed retrospectively and assigned randomly into the primary cohort (n=82) and validation cohort (n=41). Tumour proliferative activity was defined by the Ki-67 index based on immunohistochemistry. Standard uptake value (SUV) maps were generated, and 1,561 radiomics features were extracted. Radiomics features were selected through the sequential application of three algorithms. Three predictive signatures were generated: a radiomics signature, a clinical signature, and a fusion signature. The predictive performances were evaluated by receiver operating characteristic (ROC) curve analysis, and patient prognoses were stratified based on the Ki-67 index and the signature with the most reliable performance. RESULTS: Nine radiomics features were selected to construct the radiomics signature that achieved an accuracy of 81.7% and 73.2% and an area under the curve (AUC) of 0.88 and 0.76 in the primary cohort and the validation cohort, respectively. The clinical signature and fusion signature demonstrated comparable performance in the primary cohort but were over-fitted judging from the result in the validation cohort. Both the Ki-67 index and the radiomics signature could stratify patients into two distinctive prognostic groups, and the difference within each prognostic group was not statistically significant. CONCLUSION: Radiomics signature based on 18F-FDG-PET is a promising method for the non-invasive measurement of glioma proliferative activity and facilitates the prediction of patient prognoses.
AIM: To investigate a radiomics method based on 2-[18F]-fluoro-2-deoxy-d-glucose (FDG) positron-emission tomography (PET) to non-invasively evaluate proliferative activity in gliomas. MATERIALS AND METHODS: A total of 123 patients with histopathologically confirmed primary glioma were reviewed retrospectively and assigned randomly into the primary cohort (n=82) and validation cohort (n=41). Tumour proliferative activity was defined by the Ki-67 index based on immunohistochemistry. Standard uptake value (SUV) maps were generated, and 1,561 radiomics features were extracted. Radiomics features were selected through the sequential application of three algorithms. Three predictive signatures were generated: a radiomics signature, a clinical signature, and a fusion signature. The predictive performances were evaluated by receiver operating characteristic (ROC) curve analysis, and patient prognoses were stratified based on the Ki-67 index and the signature with the most reliable performance. RESULTS: Nine radiomics features were selected to construct the radiomics signature that achieved an accuracy of 81.7% and 73.2% and an area under the curve (AUC) of 0.88 and 0.76 in the primary cohort and the validation cohort, respectively. The clinical signature and fusion signature demonstrated comparable performance in the primary cohort but were over-fitted judging from the result in the validation cohort. Both the Ki-67 index and the radiomics signature could stratify patients into two distinctive prognostic groups, and the difference within each prognostic group was not statistically significant. CONCLUSION: Radiomics signature based on 18F-FDG-PET is a promising method for the non-invasive measurement of glioma proliferative activity and facilitates the prediction of patient prognoses.