PT - JOURNAL ARTICLE AU - Kentaro Kobayashi AU - Kenji Hirata AU - Shigeru Yamaguchi AU - Hiroyuki Kobayashi AU - Shunsuke Terasaka AU - Osamu Manabe AU - Takuya Toyonaga AU - Tohru Shiga AU - Nagara Tamaki TI - Texture analysis of <sup>18</sup>F-FDG PET may differentiate glioblastoma from lower grade glioma. DP - 2018 May 01 TA - Journal of Nuclear Medicine PG - 1682--1682 VI - 59 IP - supplement 1 4099 - http://jnm.snmjournals.org/content/59/supplement_1/1682.short 4100 - http://jnm.snmjournals.org/content/59/supplement_1/1682.full SO - J Nucl Med2018 May 01; 59 AB - 1682Objectives: 18F-Fluorodeoxyglucose (FDG) PET is used extensively for assessment of the histological grading of gliomas. The 2016 WHO guideline emphasizes the importance of differentiating glioblastoma (grade IV) from lower grade gliomas, i.e., grade II and III gliomas. However, both glioblastoma and grade III gliomas tend to show high FDG uptake and thus its differentiation is difficult when using conventional parameters such as maximum of standardized uptake value (SUVmax) and tumor-to-normal ratio (TNR). Since SUVmax and TNR reflect only a single voxel of the tumor, these parameters do not make use of information of intratumoral heterogeneity. Recently, in the investigation of malignancies other than brain tumor, texture features derived from FDG PET have been demonstrated to be useful in differential diagnosis and prediction of prognosis. We hypothesized that texture features may different between different grades, because grade IV more often has pathological necrosis in the tumor than grade III. Therefore, we aimed to clarify whether measurement of FDG uptake heterogeneity in glioma patients using texture analysis is useful in differential diagnosis, compared to conventional parameters. Methods: A total of 26 glioma patients underwent FDG PET with a high-resolution PET scanner (ECAT HR+ scanner, Asahi-Siemens Medical Technologies Ltd., Tokyo, Japan). The clinical characteristics of patients are summarized in Table 1. The TNR was defined as SUVmax of the lesion divided by the reference brain tissue, i.e., the contralateral frontal lobe cortex. FDG PET images were coregistered with individual Gd-enhanced T1-weighted MR images using SPM8. The volume of interest was manually delineated on MR images. The VOI voxel intensities were resampled using 64 discrete values between minimum and maximum SUVs. In addition to histogram analysis, 4 kinds of texture matrices (gray-level co-occurrence matrix (13 directions), gray-level run length matrix (13 directions), gray-level zone size matrix, and neighborhood gray-level difference matrix) were generated to calculate a total of 36 texture parameters (Figure 1). We evaluated SUVmax, TNR, and tumor volume (TV) as well as textural parameters, compared with histopathologic grading as the gold standard. Results: Of the 36 texture features, 21 parameters reached significant difference (P&lt;0.05) between glioblastoma (grade IV) vs. lower grade glioma (grade II and III, Table 2). When Bonferroni correction was performed, the significance level was p &lt;0.0013, and the five parameters showed a significant difference. Among texture features, significant level was the highest for HGRE (1081 ± 65 vs. 685.2 ±82.3, p=0.0009), with sensitivity, specificity, and accuracy being 87.5%, 90.0%, and 88.4%, respectively. None of conventional parameters (SUVmax, TNR, and TV) reached significant difference between glioblastoma vs. lower grade gliomas(table 3). Conclusions: Textural analysis of FDG PET may be useful for grading gliomas, reflecting intratumoral glucose metabolic heterogeneity. That cannot be assessed by conventional parameters simply expressing intensity of glucose metabolism.