MP74-15 MACHINE LEARNING-BASED CONSTRUCTION AND VALIDATION OF A 68 GA-PSMA PET/CT RADIOMICS MODEL FOR PREDICTING ISUP GRADING IN PROSTATE CANCER

You have accessJournal of UrologyProstate Cancer: Detection & Screening VI (MP74), Moderated Poster 741 May 2024MP74-15 MACHINE LEARNING-BASED CONSTRUCTION AND VALIDATION OF A 68GA-PSMA PET/CT RADIOMICS MODEL FOR PREDICTING ISUP GRADING IN PROSTATE CANCER Honghu Zhang, Yongxiang Tang, Lin Qi, Minfeng Chen, Xiaomei Gao, Shuo Hu, and Yi Cai Honghu ZhangHonghu Zhang , Yongxiang TangYongxiang Tang , Lin QiLin Qi , Minfeng ChenMinfeng Chen , Xiaomei GaoXiaomei Gao , Shuo HuShuo Hu , and Yi CaiYi Cai View All Author Informationhttps://doi.org/10.1097/01.JU.0001008632.59099.b9.15AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: The International Society of Urological Pathology (ISUP) grading of prostate cancer is one of the key determinants in the management and treatment planning for PCa patients. Accurate non-invasive assessment of the ISUP grading group would be highly beneficial for decisions regarding biopsy targeting and treatment. The use of PSMA-PET/CT radiomics for predicting ISUP has not been widely studied. The aim of this study is to investigate the role of 68 Ga-PSMA-11 PET/CT radiomics in predicting the ISUP grading of primary prostate cancer (PCa). METHODS: This retrospective study included 252 prostate cancer patients who underwent 68Ga-PSMA-11 PET/CT before prostate biopsy or radical prostatectomy. The entire prostate was used as the volume of interest (VOI). Significantly discriminative features were selected using the classical Minimum Redundancy Maximum Relevance (mRMR) algorithm and the Least Absolute Shrinkage and Selection Operator (LASSO) algorithm. In each subset of analysis, 8 separate machine learning (ML) classifiers were trained and tested using ten-fold cross-validation. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and the area under the receiver operating characteristic curve (AUC) were calculated for each model. RESULTS: Among the 252 patients, 99 had prostate cancer with ISUP <4, and 153 had prostate cancer with ISUP ≥4. In the best-performing prediction model in the test cohort, the group AUC was 0.834 (Sensitivity, 75.6%, specificity, 82.8%, PPV, 0.872, and NPV, 0.686). Furthermore, when combining PSA and SUVmax, a more robust prediction model was constructed, with a training group AUC of 0.877(Sensitivity, 75.6%, specificity, 89.7%, PPV, 0.919, and NPV, 0.703). CONCLUSIONS: This study established and validated a radiomics model based on 68Ga-PSMA PET/CT, providing an accurate and non-invasive method for predicting ISUP grading. Additionally, it opened up new avenues for combining radiomics with clinical pathology data to predict ISUP grading. Source of Funding: This research was supported by the key Research and Development program of Hunan Province (2021SK2014, 2023SK2017), the Science and Technology Innovation Team Talent Project of Hunan Province (2021RC4056), the National Natural Science Foundation of China(82272907, 81974397, 91859207, 81771873), and the clinical research foundation of the National Clinical Research Center for Geriatric Diseases (XIANGYA; 2020LNJJ01, 2022LNJJ13) © 2024 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 211Issue 5SMay 2024Page: e1199 Advertisement Copyright & Permissions© 2024 by American Urological Association Education and Research, Inc.Metrics Author Information Honghu Zhang More articles by this author Yongxiang Tang More articles by this author Lin Qi More articles by this author Minfeng Chen More articles by this author Xiaomei Gao More articles by this author Shuo Hu More articles by this author Yi Cai More articles by this author Expand All Advertisement PDF downloadLoading ...