Mp35-01 prostate tumor textural heterogeneity of 11c-acetate positron emission tomography and t2-weighted magnetic resonance imaging correlate with biochemical recurrence: preliminary findings

You have accessJournal of UrologyProstate Cancer: Markers I1 Apr 2018MP35-01 PROSTATE TUMOR TEXTURAL HETEROGENEITY OF 11C-ACETATE POSITRON EMISSION TOMOGRAPHY AND T2-WEIGHTED MAGNETIC RESONANCE IMAGING CORRELATE WITH BIOCHEMICAL RECURRENCE: PRELIMINARY FINDINGS Lin Li, Ivan Jambor, Pekka Taimen, Harri Merisaar, Heikki Minn, Peter Bostrom, Hannu Aronen, Ahmad Algohary, and Anant Madabhushi Lin LiLin Li More articles by this author , Ivan JamborIvan Jambor More articles by this author , Pekka TaimenPekka Taimen More articles by this author , Harri MerisaarHarri Merisaar More articles by this author , Heikki MinnHeikki Minn More articles by this author , Peter BostromPeter Bostrom More articles by this author , Hannu AronenHannu Aronen More articles by this author , Ahmad AlgoharyAhmad Algohary More articles by this author , and Anant MadabhushiAnant Madabhushi More articles by this author View All Author Informationhttps://doi.org/10.1016/j.juro.2018.02.1114AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookTwitterLinked InEmail INTRODUCTION AND OBJECTIVES Multi-parametric prostate magnetic resonance imaging (MRI) has been primarily used for prostate cancer (PCa) localization and recently for Gleason score prediction. 11C-acetate positron emission tomography (PET) is a promising imaging modality for detection of metastatic PCa. Computerized tomography (CT) is commonly acquired along with PET for attenuation correction but typically provides limited anatomical information. Recently, there has been interest in the use of radiomics or computer extracted image features to characterize tumor heterogeneity patterns. Hence, we sought to evaluate the role of radiomics to characterize and associate tumor heterogeneity patterns within MRI, 11C-PET and CT to differentiate patients with (BCR+) and without (BCR-) biochemical recurrence (BCR) following surgery. METHODS T2-weighted imaging (T2w), dynamic contrast enhanced MRI (DCE-MRI) and 11C-PET/CT data sets of 10 (5 with BCR+) patients with at least 8 years of clinical follow up after prostatectomy were included. The tumor annotations from whole-mount prostatectomy specimens, as 'gold standard', were registered to T2w, DCE-MRI and PET/CT. We assessed 180 texture features for each imaging modality within tumor ROIs and calculated first order statistics. Pearson correlation (R < 0.5) was used to reduce redundant features. Wilcoxon rank sum test was adopted to select representative features for BCR (p-value < 0.05). In addition, variable importance on projection (VIP) measure for principle component analysis (PCA) was used to rank order features in terms of discriminability. RESULTS Figure 1 shows the unsupervised consensus clustering results for discriminating BCR+ and BCR- patients based on the top 2 PCA-VIP ranked features. These results suggest that textural heterogeneity of PET and T2w MRI potentially correlate with BCR, with increased textural expression (and hence heterogeneity) within the tumor on both PET and T2w MRI being reflected in the BCR+ patients. CONCLUSIONS This preliminary study suggests that (1) texture analysis enables characterization of heterogeneity patterns within multimodal imaging and (2) heterogeneity patterns from PET and T2w MRI appeared to be associated with post-surgical BCR. © 2018FiguresReferencesRelatedDetails Volume 199Issue 4SApril 2018Page: e446 Advertisement Copyright & Permissions© 2018MetricsAuthor Information Lin Li More articles by this author Ivan Jambor More articles by this author Pekka Taimen More articles by this author Harri Merisaar More articles by this author Heikki Minn More articles by this author Peter Bostrom More articles by this author Hannu Aronen More articles by this author Ahmad Algohary More articles by this author Anant Madabhushi More articles by this author Expand All Advertisement Advertisement PDF downloadLoading ...