A new assessment model for tumor heterogeneity analysis with [18]F-FDG PET images

Authors

  • Ping Wang Department of Radiation Oncology, Tianjin Medical University Cancer Institute and Hospital, Tianjin, China
  • Wengui Xu Department of Molecular Imaging and Nuclear Medicine, Tianjin Medical University Cancer Institute and Hospital, Tianjin, China
  • Jian Sun Department of Radiation Oncology, Tianjin Medical University Cancer Institute and Hospital, Tianjin, China
  • Chengwen Yang Department of Radiation Oncology, Tianjin Medical University Cancer Institute and Hospital, Tianjin, China; Department of Biomedical Engineering, Tianjin University, Tianjin, China
  • Gang Wang Department of Biomedical Engineering, Tianjin University, Tianjin, China
  • Yu Sa Department of Biomedical Engineering, Tianjin University, Tianjin, China
  • Xin-Hua Hu Department of Biomedical Engineering, Tianjin University, Tianjin, China; Department of Physics, East Carolina University, Greenville, NC
  • Yuanming Feng Department of Radiation Oncology, Tianjin Medical University Cancer Institute and Hospital, Tianjin, China; Department of Biomedical Engineering, Tianjin University, Tianjin, China

DOI:

https://doi.org/10.17179/excli2015-723

Keywords:

[18]F-FDG PET image, standard uptake value, intratumor heterogeneity, gray level co-occurrence matrix

Abstract

It has been shown that the intratumor heterogeneity can be characterized with quantitative analysis of the [18]F-FDG PET image data. The existing models employ multiple parameters for feature extraction which makes it difficult to implement in clinical settings for the quantitative characterization. This article reports an easy-to-use and differential SUV based model for quantitative assessment of the intratumor heterogeneity from 3D [18]F-FDG PET image data. An H index is defined to assess tumor heterogeneity by summing voxel-wise distribution of differential SUV from the [18]F-FDG PET image data. The summation is weighted by the distance of SUV difference among neighboring voxels from the center of the tumor and can thus yield increased values for tumors with peripheral sub-regions of high SUV that often serves as an indicator of augmented malignancy. Furthermore, the sign of H index is used to differentiate the rate of change for volume averaged SUV from its center to periphery. The new model with the H index has been compared with a widely-used model of gray level co-occurrence matrix (GLCM) for image texture characterization with phantoms of different configurations and the [18]F-FDG PET image data of 6 lung cancer patients to evaluate its effectiveness and feasibility for clinical uses. The comparison of the H index and GLCM parameters with the phantoms demonstrate that the H index can characterize the SUV heterogeneity in all of 6 2D phantoms while only 1 GLCM parameter can do for 1 and fail to differentiate for other 2D phantoms. For the 8 3D phantoms, the H index can clearly differentiate all of them while the 4 GLCM parameters provide complicated patterns in the characterization. Feasibility study with the PET image data from 6 lung cancer patients show that the H index provides an effective single-parameter metric to characterize tumor heterogeneity in terms of the local SUV variation, and it has higher correlation with tumor volume change after radiotherapy (R2 = 0.83) than the 4 GLCM parameters (R2 = 0.63, 0.73, 0.59 and 0.75 for Energy, Contrast, Local Homogeneity and Entropy respectively). The new model of the H index has the capacity to characterize the intratumor heterogeneity feature from 3D [18]F-FDG PET image data. As a single parameter with an intuitive definition, the H index offers potential for clinical applications.

Downloads

Published

2016-01-28

How to Cite

Wang, P., Xu, W., Sun, J., Yang, C., Wang, G., Sa, Y., … Feng, Y. (2016). A new assessment model for tumor heterogeneity analysis with [18]F-FDG PET images. EXCLI Journal, 15, 75–84. https://doi.org/10.17179/excli2015-723

Issue

Vol. 15 (2016)

Section

Original articles

License

Authors who publish in this journal agree to the following terms:

  • The authors keep the copyright and grant the journal the right of first publication under the terms of the Creative Commons Attribution license, CC BY 4.0. This licencse permits unrestricted use, distribution and reproduction in any medium, provided that the original work is properly cited.
  • The use of general descriptive names, trade names, trademarks, and so forth in this publication, even if not specifically identified, does not imply that these names are not protected by the relevant laws and regulations.
  • Because the advice and information in this journal are believed to be true and accurate at the time of publication, neither the authors, the editors, nor the publisher accept any legal responsibility for any errors or omissions presented in the publication. The publisher makes no guarantee, express or implied, with respect to the material contained herein.
  • The authors can enter into additional contracts for the non-exclusive distribution of the journal's published version by citing the initial publication in this journal (e.g. publishing in an institutional repository or in a book).