PD41-02 RADIOMIC-BASED “VIRTUAL BIOPSY” RELIABLY DIFFERENTIATES BENIGN FROM MALIGNANT RENAL MASSES

Abstract

You have accessJournal of UrologySurgical Technology & Simulation: Instrumentation & Technology II (PD41)1 Sep 2021PD41-02 RADIOMIC-BASED “VIRTUAL BIOPSY” RELIABLY DIFFERENTIATES BENIGN FROM MALIGNANT RENAL MASSES Nima Nassiri, Marissa Maas, Giovanni Cacciamani, Bino Varghese, Darryl Hwang, Assad Oberai, Xiaomeng Lei, Mihir Desai, Monish Aron, Steven Cen, Vinay Duddalwar, and Inderbir Gill Nima NassiriNima Nassiri More articles by this author , Marissa MaasMarissa Maas More articles by this author , Giovanni CacciamaniGiovanni Cacciamani More articles by this author , Bino VargheseBino Varghese More articles by this author , Darryl HwangDarryl Hwang More articles by this author , Assad OberaiAssad Oberai More articles by this author , Xiaomeng LeiXiaomeng Lei More articles by this author , Mihir DesaiMihir Desai More articles by this author , Monish AronMonish Aron More articles by this author , Steven CenSteven Cen More articles by this author , Vinay DuddalwarVinay Duddalwar More articles by this author , and Inderbir GillInderbir Gill More articles by this author View All Author Informationhttps://doi.org/10.1097/JU.0000000000002051.02AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: A substantial proportion of patients undergo treatment for renal masses where active surveillance or observation may be more appropriate. We sought to determine whether a radiomic-based artificial intelligence (AI) platform that incorporates clinical variables can accurately differentiate benign renal masses from renal cell carcinoma (RCC). METHODS: We performed a cross-sectional analysis of a prospectively collected and maintained registry of patients with renal masses in accordance with the STROBE Statement for observational studies. All participants had a clinically-localized renal mass visualized with multiphase computed-tomography (CT) before robotic, laparoscopic, or open partial or radical nephrectomy. Radiomic analysis of pre-operative CT scans, specifically CT shape and texture analysis, was performed. 10-fold cross-validation was performed to overcome heterogeneities in radiomic analysis. Clinical and radiomic variables of interest (VOIs) were identified through decision tree analysis and incorporated into machine-learning Random Forest and REAL Adaboost predictive models. The primary outcome was the diagnostic capacity of the predictive model to discriminate benign renal masses from RCC. Secondary outcomes included a subanalysis for small renal masses (SRMs) and for patients who underwent percutaneous renal mass biopsy. RESULTS: 684 patients with a renal mass identified on CT imaging underwent radical or partial nephrectomy. Of the patients, 76% had RCC; 57% had an SRM, of which 73% were malignant. REAL Adaboost predictive modeling differentiated benign pathology from RCC with an AUC of 0.84 (95% C.I. 0.79-0.9) (Figure 1A). In SRMs, radiomic analysis yielded a discriminatory AUC of 0.77 (95%. C.I. 0.69-0.85) (Figure 1B). When negative and non-diagnostic biopsies were supplemented with the results of radiomic analysis, accuracy increased from 83.3% to 93.4%. CONCLUSIONS: Radiomic-based predictive modeling produces a “virtual biopsy” that may discriminate benign renal masses from RCC. Enhanced diagnostic predictability may preclude surgical intervention and increase the utility of active surveillance protocols. Source of Funding: no disclosures © 2021 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 206Issue Supplement 3September 2021Page: e680-e681 Advertisement Copyright & Permissions© 2021 by American Urological Association Education and Research, Inc.MetricsAuthor Information Nima Nassiri More articles by this author Marissa Maas More articles by this author Giovanni Cacciamani More articles by this author Bino Varghese More articles by this author Darryl Hwang More articles by this author Assad Oberai More articles by this author Xiaomeng Lei More articles by this author Mihir Desai More articles by this author Monish Aron More articles by this author Steven Cen More articles by this author Vinay Duddalwar More articles by this author Inderbir Gill More articles by this author Expand All Advertisement Loading ...