Objective To evaluate whether high-frequency ultrasound (HF-US) and ultrasound-guided core needle biopsy (CNB) can accurately detect muscle-invasive bladder cancer (MIBC) in an ex-vivo setting. Methods In this prospective study, patients undergoing radical cystectomy (RC) for urothelial carcinoma underwent ex-vivo HF-US of the extirpated bladder to assess depth of tumor invasion (ultrasound stage, usT). In a subset, HF-US guided CNB of the opened bladder targeted the deepest region of tumor invasion (cnbT). Ultrasound images and biopsy specimens were interpreted by a blinded GU-radiologist and pathologist, respectively and were compared with RC pathologic stage (pT). Co-primary endpoints were accuracy of HF-US and CNB in identifying MIBC. Results Among 54 patients, 48 underwent HF-US and 32 underwent CNB. HF-US demonstrated an accuracy of 89.6% for detecting MIBC, with sensitivity 95.0%, specificity 85.7%, and negative predictive value (NPV) 96.0%. CNB accuracy was 90.6%, with specificity and positive predictive value of 100%. In patients undergoing both modalities (n=25), a combined approach demonstrated 92.0% accuracy, 100% sensitivity, and 100% NPV. Conventional clinical staging had significantly lower accuracy (55.6%), particularly following neoadjuvant systemic therapy (40.0%). Diagnostic performance of HF-US and CNB remained high in patients treated with neoadjuvant systemic or intravesical therapy. Conclusion HF-US and ultrasound-guided CNB were highly accurate in determining MIBC and specifically demonstrated excellent NPV. These findings support further investigation of in-vivo endoscopic ultrasound and CNB staging to improve risk stratification and treatment selection in bladder cancer.

While computed tomography (CT) is the preferred imaging modality for kidney stone detection and measurement, quantifying the accuracy and precision of stone size metrics is needed, particularly with the growing use of automated image analysis tools to measure kidney stone size. A phantom study was conducted using 120 kidney stones of various known maximum diameters (ranging from 1.4 to 9.9 mm) and compositions (60 calcium, 30 uric acid, and 30 mixed calcium and uric acid). The stones were placed in an anthropomorphic phantom and CT images were reconstructed with slice thicknesses of 1, 3, or 5 mm and processed using an in-house quantitative Stone Analysis Software (qSAS) to detect each stone and determine its maximum diameter. Separately, we also assessed the agreement between qSAS and a radiologist in determining the maximum diameter of 45 kidney stones on CT images (1-mm slice) from patients. Stone detection frequencies by qSAS for stones in phantoms were 100

Automated deep learning–based segmentation is increasingly used in medical imaging to enable rapid biomarker extraction. Manual quality control (QC) of segmentation outputs remains standard prior to downstream analysis. In autosomal dominant polycystic kidney disease (ADPKD), reliable segmentation is essential for accurate total kidney volume measurement, a key biomarker for disease monitoring. We developed and evaluated a 3D deep learning model that classified MR volumes and their automated segmentation outputs into downstream Accept, Reject, and Rework categories. Criteria included scan quality, native kidney coverage, and segmentation accuracy. We used a patient-disjoint dataset of 10,749 abdominal MR scans (T2-weighted coronal series) across 5708 exams from 2717 patients with ADPKD for model development. Model training utilized 9549 scans, and performance was validated on a balanced set of 600 scans against manual labels. A final balanced holdout test set of 600 scans was used for evaluation. Across three reader-defined reference standards, DenseNet121 achieved macro F1-scores of 0.78, 0.80, and 0.86 with accuracies of 79

BACKGROUND:Total kidney and liver volumes are key image-based biomarkers to predict the severity of kidney and liver phenotype in autosomal dominant polycystic kidney disease (ADPKD). However, MRI-based advanced biomarkers like total cyst number (TCN) and cyst parenchyma surface area (CPSA) have been shown to more accurately assess cyst burden and improve the prediction of disease progression. The main aim of this study is to extend the calculation of advanced biomarkers to other imaging modalities; thus, we propose a fully automated model to segment kidney and liver cysts in CT images. METHODS:Abdominal CTs of ADPKD patients were gathered retrospectively between 2001-2018. A 3D deep-learning method using the nnU-Net architecture was trained to learn cyst edges-cores and the non-cystic kidney/liver parenchyma. Separate segmentation models were trained for kidney cysts in contrast-enhanced CTs and liver cysts in non-contrast CTs using an active learning approach. Two experienced research fellows manually generated the reference standard segmentation, which were reviewed by an expert radiologist for accuracy. RESULTS:Two-hundred CT scans from 148 patients (mean age, 51.2 ± 14.1 years; 48% male) were utilized for model training (80%) and testing (20%). In the test set, both models showed good agreement with the reference standard segmentations, similar to the agreement between two independent human readers (model vs reader: TCNkidney/liver r=0.96/0.97 and CPSAkidney r=0.98), inter-reader: TCNkidney/liver r=0.96/0.98 and CPSAkidney r=0.99). CONCLUSIONS:Our study demonstrates that automated models can segment kidney and liver cysts accurately in CT scans of patients with ADPKD.

Integration of AI-enabled algorithms into the radiology workflow presents a complex array of challenges that span operational, technical, clinical, and regulatory domains. Successfully overcoming these hurdles requires a multifaceted approach, including strategic planning, educational initiatives, and careful consideration of the practical implications for radiologists' workloads. Institutions must navigate these challenges with a clear understanding of the potential benefits and limitations of both vended and in-house developed AI tools.

Background: During surgical extirpation of cystic renal masses, surgeons attempt to avoid cyst rupture due to the theoretical risk of tumor seeding. Whether the concern regarding tumor seeding is warranted is debatable. Our objective was to evaluate the presence of malignant cells in the fluid of complex renal cysts. Methods: This was a cross-sectional analysis of adult patients undergoing radical or partial nephrectomy to address a cystic renal mass. Patients undergoing a partial or radical nephrectomy by open or robotic approach for a clinically localized (< cT2N0M0) cystic renal mass were included. Following excision, fluid from the mass was aspirated and sent for cytologic analysis. Cyst fluid was prepared by processing up to 50 mL into a PreservCyt (R) vial on a ThinPrep (R) 2000 or ThinPrep (R) 5000 processor using standard protocols, resulting in a pap-stained ThinPrep glass slide. The second half of the fluid was processed into a cellblock using a plasma/thrombin process resulting in a Formalin-Fixed Paraffin-Embedded (FFPE) block cut to produce a hematoxylin and eosin (H&E)-stained slide. Both the pap-stained and H&E slides were evaluated for malignant cells by a cytotechnologist and pathologist. Results: Twenty-three patients underwent resection of 24 cystic tumors including 17 (73.9%) males and 6 (26.1%) females. The median patient age was 58 years [interquartile range (IQR), 43-68 years]. The median tumor diameter was 3.7 cm (IQR, 3.2-6.1 cm). Most patients underwent robotic partial nephrectomy (n=19, 83%). Renal cyst cytology was benign in 46% (n=11), atypical in 29% (n=7), suspicious in 8% (n=2), positive for neoplasm in 4% (n=1), and positive for malignancy in 4% (n=1). Clear cell renal cell carcinoma was the most common histologic subtype (n=17, 71%). Conclusions: Based on routine cytologic analysis, there is no clear pattern with the presence or absence of malignant cells in the fluid of complex renal cysts. More sophisticated testing may provide insight into the malignant potential of renal cyst fluid.

To compare same-day photon-counting detector CT (PCD-CT) to conventional energy-integrating detector CT (EID-CT) for detection of small renal stones (≤ 3 mm). Patients undergoing clinical dual-energy EID-CT for known or suspected stone disease underwent same-day research PCD-CT. Patients with greater than 10 stones and no visible stones under 3 mm were excluded. Three radiologists selected the optimal reconstruction configuration for each CT modality and created the reference standard for renal stone presence. Two other radiologists, blinded to imaging modality, independently reviewed anonymized images to detect renal stones, rating confidence in potential stones using a Likert scale (1 = Definitely present, 2 = Probably present, 3 = Questionably present, 4 = Not seen). Sensitivity and false positive detections for PCD and EID-CT were calculated. Twenty-one patients underwent clinical EID-CT followed by same-day PCD-CT, with the reference standard identifying 121 renal stones (mean size 2.8 ± 2.6 mm). 0.4-mm PCD-CT images were more likely to display a stone as definitely present compared to 1- or 2-mm EID-CT images (p < 0.0001). Overall sensitivity for detection of all stones was greater at PCD-CT (0.75 vs. 0.55, p < 0.05). Pooled sensitivity of stones ≤ 3 mm was also significantly higher at PCD-CT (0.67 vs. 0.41, p < 0.05), with false positive detections differing between readers and modalities (PCD-CT vs. EID-CT: R1—7 v. 5; R2 – 7 v. 1). Sensitivity for renal stones was significantly higher using high spatial resolution PCD-CT vs. EID-CT, especially for stones 3 mm or less in size, which may be important for at-risk patient populations. Prospective evaluation in larger patient populations that will benefit from detection of small stones is warranted.

In patients with renal masses, when intervention is warranted, partial nephrectomy is preferred when feasible, especially for T1 renal masses. Thermal ablation, however, has become an accepted alternative treatment of small renal neoplasms with excellent oncologic outcomes. The National Comprehensive Cancer Network guidelines include thermal ablation as a treatment strategy for managing T1a and select T1b masses. Other potential locoregional treatments for malignant and benign renal masses are emerging. Stereotactic body radiation therapy (SBRT) is gaining traction in the management of select localized renal masses and can be considered in patients who are not surgical candidates. Imaging findings after SBRT differ from those after thermal ablation. As this technique becomes more widely used, the temporal evolution of postSBRT changes needs to be understood to correctly identify local tumor progression. Renal artery embolization is accepted as safe and effective for patients with renal angiomyolipoma. It can also be used in the management of renal cell carcinoma as an adjunct to definitive treatment or for palliation in advanced disease. With the growing acceptance of local-regional treatment of renal masses and the corresponding number of related patients undergoing surveillance and postablation imaging, the diagnostic radiologist will have greater responsibility in the appropriate interpretation of follow-up imaging to accurately assess normal postprocedural findings and define oncologic outcomes. Such interpretations should be based on understanding of the procedure and related imaging findings, both acutely and in the years after treatment. Procedural complications and recurrent tumor are identifiable based on deviations from the expected course. (c) RSNA, 2024 center dot radiographics.rsna.org

PURPOSE:Renal cell carcinoma (RCC) is known to form venous tumor thrombus (VTT); however, limited data exist on VTT growth rates. We sought to characterize the growth rate of VTT from RCC and identify clinicopathologic features associated with faster thrombus growth. MATERIALS AND METHODS:Systemic therapy-naïve patients with RCC and VTT with 2 preoperative imaging studies performed with contrast at least 7 days apart were identified from a single institutional registry. VTT levels were assigned using the Mayo classification and measured in mm. Changes between scans were linearized to growth rate per day. Multivariable regression models identified factors associated with VTT growth. RESULTS:We evaluated 141 patients with a median time between scans of 20 days (IQR: 12-30) and noted a median VTT growth rate of 0.3 mm/d. Sixteen (11%) patients were reclassified on the second scan to a higher Mayo VTT level. There was a significant relationship between VTT growth and increasing tumor thrombus level (P < .001) as well as the presence of sarcomatoid/rhabdoid features (0.4 vs 0.3 mm/d, P = .003). On multivariable regression, there is a quadratic increase in VTT growth rates as thrombus level increases (P = .001). Sarcomatoid/rhabdoid features (coefficient +0.43 mm/d, P < .001) and metastatic disease (coefficient +0.30 mm/d, P = .03) were also associated with increased tumor thrombus growth rates. CONCLUSIONS:VTT growth is generally less than 1 mm/d, although higher level thrombi and disease with aggressive histopathology grew more rapidly. These data improve the understanding of tumor thrombus growth and can aid in preoperative triage and surgical planning.

BackgroundNational guidelines recommend that waist circumference (WC) be measured in patients with a body mass index (BMI) 27–35 kg/m2. Unfortunately, perhaps due to logistical reasons, WC is seldom measured in clinical settings. Herein, we describe the performance of a self-operated waist measurement device (SOWMD) as a potential means to overcome barriers to measuring WC.Materials and methodsTen volunteers underwent WC measures by professionals and SOWMD on 5 separate days to assess the reproducibility and accuracy. We then compared SOWMD measures with CT-derived fat content by recruiting 81 patients scheduled for a diagnostic abdominal CT scan.ResultsThere was no difference between professionally measured and SOWMD-measured WC; the intraindividual coefficient of variation over the 5 days was between 0.4% and 2.2%. The WC measured manually, by SOWMD and CT scan were highly correlated (r=0.90–0.92, all p<0.001). The minimal WC measured by SOWMD was a better predictor (r=0.81 for all patients, r=0.87 for men, both p<0.001) of CT-measured visceral adipose tissue volume than other approaches. The minimal WC measured by SOWMD was correlated with fasting plasma glucose (r=0.40, p<0.05), triglyceride (r=0.41, p<0.01) and high-density cholesterol (r=−0.49, p<0.001) concentrations.ConclusionSOWMD is a reproducible, accurate and convenient way to measure WC that can provide added value for health care providers when combined with BMI information.

RATIONALE & OBJECTIVE:Simple kidney cysts, which are common and usually considered of limited clinical relevance, are associated with older age and lower glomerular filtration rate (GFR), but little has been known of their association with progressive chronic kidney disease (CKD). STUDY DESIGN:Observational cohort study. SETTING & PARTICIPANTS:Patients with presurgical computed tomography or magnetic resonance imaging who underwent a radical nephrectomy for a tumor; we reviewed the retained kidney images to characterize parenchymal cysts at least 5mm in diameter according to size and location. EXPOSURE:Parenchymal cysts at least 5mm in diameter in the retained kidney. Cyst characteristics were correlated with microstructural findings on kidney histology. OUTCOME:Progressive CKD defined by dialysis, kidney transplantation, a sustained≥40% decline in eGFR for at least 3 months, or an eGFR<10mL/min/1.73m2 that was at least 5mL/min/1.73m2 below the postnephrectomy baseline for at least 3 months. ANALYTICAL APPROACH:Cox models assessed the risk of progressive CKD. Models adjusted for baseline age, sex, body mass index, hypertension, diabetes, eGFR, proteinuria, and tumor volume. Nonparametric Spearman's correlations were used to examine the association of the number and size of the cysts with clinical characteristics, kidney function, and kidney volumes. RESULTS:There were 1,195 patients with 50 progressive CKD events over a median 4.4 years of follow-up evaluation. On baseline imaging, 38% had at least 1 cyst, 34% had at least 1 cortical cyst, and 8.7% had at least 1 medullary cyst. A higher number of cysts was associated with progressive CKD and was modestly correlated with larger nephrons and more nephrosclerosis on kidney histology. The number of medullary cysts was more strongly associated with progressive CKD than the number of cortical cysts. LIMITATIONS:Patients who undergo a radical nephrectomy may differ from the general population. A radical nephrectomy may accelerate the risk of progressive CKD. Genetic testing was not performed. CONCLUSIONS:Cysts in the kidney, particularly the medulla, should be further examined as a potentially useful imaging biomarker of progressive CKD beyond the current clinical evaluation of kidney function and common CKD risk factors. PLAIN-LANGUAGE SUMMARY:Kidney cysts are common and often are considered of limited clinical relevance despite being associated with lower glomerular filtration rate. We studied a large cohort of patients who had a kidney removed due to a tumor to determine whether cysts in the retained kidney were associated with kidney health in the future. We found that more cysts in the kidney and, in particular, cysts in the deepest tissue of the kidney (the medulla) were associated with progressive kidney disease, including kidney failure where dialysis or a kidney transplantation is needed. Patients with cysts in the kidney medulla may benefit from closer monitoring.

The lymphatic system (or lymphatics) consists of lymphoid organs and lymphatic vessels. Despite the numerous previously published studies describing conditions related to perirenal and intrarenal lymphoid organs in the radiology literature, the radiologic findings of conditions related to intrarenal and perirenal lymphatic vessels have been scarcely reported. In the renal cortex, interlobular lymphatic capillaries do not have valves; therefore, lymph can travel along the primary route toward the hilum, as well as toward the capsular lymphatic plexus. These two lymphatic pathways can be opacified by contrast medium via pyelolymphatic backflow at CT urography, which reflects urinary contrast agent leakage into perirenal lymphatic vessels via forniceal rupture. Pyelolymphatic backflow toward the renal hilum should be distinguished from urinary leakage due to urinary injury. Delayed subcapsular contrast material retention via pyelolymphatic backflow, appearing as hyperattenuating subcapsular foci on CT images, mimics other subcapsular cystic diseases. In contrast to renal parapelvic cysts originating from the renal parenchyma, renal peripelvic cysts are known to be of lymphatic origin. Congenital renal lymphangiectasia is mainly seen in children and assessed and followed up at imaging. Several lymphatic conditions, including lymphatic leakage as an early complication and acquired renal lymphangiectasia as a late complication, are sometimes identified at imaging follow-up of kidney transplant. Lymphangiographic contrast material accumulation in the renal hilar lymphatic vessels is characteristic of chylo-urinary fistula. Chyluria appears as a fat-layering fluid-fluid level in the urinary bladder or upper urinary tract. Recognition of the anatomic pathway of tumor spread via lymphatic vessels at imaging is of clinical importance for accurate management at oncologic imaging. ©RSNA, 2024 Test Your Knowledge questions for this article are available in the supplemental material.

You have accessJournal of UrologyImaging/Uroradiology II (MP30)1 May 2024MP30-07 AUTOMATED RENAL VOLUME MEASUREMENT USING ARTIFICIAL INTELLIGENCE: CORRELATION TO POST-OPERATIVE RENAL FUNCTION AFTER RADICAL AND PARTIAL NEPHRECTOMY Abhinav Khanna, Vidit Sharma, Ekamjit S. Deol, Adriana Gregory, Harrison C. Gottlich, Cole Cook, Jason Klug, Christine Lohse, Theodora Potretzke, Aaron Potretzke, Stephen A. Boorjian, R. Houston Thompson, Andrew Rule, Naoki Takahashi, Alexander Denic, Bradley Erickson, Timothy Kline, and Bradley Leibovich Abhinav KhannaAbhinav Khanna , Vidit SharmaVidit Sharma , Ekamjit S. DeolEkamjit S. Deol , Adriana GregoryAdriana Gregory , Harrison C. GottlichHarrison C. Gottlich , Cole CookCole Cook , Jason KlugJason Klug , Christine LohseChristine Lohse , Theodora PotretzkeTheodora Potretzke , Aaron PotretzkeAaron Potretzke , Stephen A. BoorjianStephen A. Boorjian , R. Houston ThompsonR. Houston Thompson , Andrew RuleAndrew Rule , Naoki TakahashiNaoki Takahashi , Alexander DenicAlexander Denic , Bradley EricksonBradley Erickson , Timothy KlineTimothy Kline , and Bradley LeibovichBradley Leibovich View All Author Informationhttps://doi.org/10.1097/01.JU.0001009416.90901.7b.07AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: Post-operative renal function (PORF) after radical nephrectomy (RN) and partial nephrectomy (PN) is correlated to the volume of parenchyma spared during surgery. However, the calculation of non-neoplastic renal volume on imaging is resource-intensive and does not translate readily into clinical practice. We examine the utility of a deep learning algorithm for automated renal volume calculation in predicting PORF following PN and RN. METHODS: We identified patients undergoing RN or PN at our tertiary referral center with accessible pre-operative CT images. We developed a novel deep learning algorithm using nnU-Net architecture to automatically measure ipsilateral renal volume (RV), contralateral RV, and kidney tumor volume on contrast-enhanced CT (Figure 1). The associations between algorithm-generated pre-operative non-neoplastic RV and observed PORF were assessed using generalized linear mixed effect models, adjusted for known clinical factors associated with PORF (age, diabetes, preoperative eGFR, proteinuria, tumor size, time from surgery). RESULTS: CT images from 1,077 patients, including 300 RN and 777 PN, were included. Mean (SD) contralateral RV as a split percentage was 53% (7) for RN and 50% (3) for PN. Mean (SD) contralateral RV as an absolute volume was 197 mL (52) for RN and 198 mL (51) for PN. Mean duration of follow-up from surgery was 45 (35) months for 3,073 postoperative eGFR assessments following RN and 50 (38) months for 7,478 postoperative eGFR assessments following PN. The mean (SD) number of postoperative eGFR measurements was 10 (9) per patient. When added to previously validated multivariable clinical models to predict PORF, higher AI-derived contralateral RV was independently associated with better PORF in RN (p<0.001) and PN (p<0.001). Each 10% increase in split contralateral RV was associated with a 6.5 and a 2.6 mL/min/1.73 m2 increase in PORF following RN and PN, respectively. CONCLUSIONS: Higher pre-operative non-neoplastic RV is associated with improved long-term renal function following RN and PN, even after adjusting for a previously validated comprehensive clinical prediction model. We developed an AI tool to automate measurement of non-neoplastic RV, which may facilitate integration of RV measurement into clinical practice. Download PPT Source of Funding: None © 2024 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 211Issue 5SMay 2024Page: e493 Advertisement Copyright & Permissions© 2024 by American Urological Association Education and Research, Inc.Metrics Author Information Abhinav Khanna More articles by this author Vidit Sharma More articles by this author Ekamjit S. Deol More articles by this author Adriana Gregory More articles by this author Harrison C. Gottlich More articles by this author Cole Cook More articles by this author Jason Klug More articles by this author Christine Lohse More articles by this author Theodora Potretzke More articles by this author Aaron Potretzke More articles by this author Stephen A. Boorjian More articles by this author R. Houston Thompson More articles by this author Andrew Rule More articles by this author Naoki Takahashi More articles by this author Alexander Denic More articles by this author Bradley Erickson More articles by this author Timothy Kline More articles by this author Bradley Leibovich More articles by this author Expand All Advertisement PDF downloadLoading ...

A 7-Year-Old Boy with Fever and Dark UrineA 7-year-old boy with surgically repaired tetralogy of Fallot presented for evaluation of fever and dark urine. How do you approach the evaluation, and what is the diagnosis?

You have accessJournal of UrologyCME1 Apr 2023PD08-04 VIRTUAL RENAL MASS BIOPSY: PREDICTING RENAL TUMOR HISTOLOGY ON ABDOMINAL CT IMAGES USING MACHINE LEARNING Abhinav Khanna, Vidit Sharma, Adriana Gregory, H. Chase Gottlich, Cole J. Cook, Jason Klug, Christine Lohse, Theodora Potretzke, Aaron Potretzke, Stephen A. Boorjian, R. Houston Thompson, Naoki Takahashi, Bradley Erickson, John Cheville, Timothy Kline, and Bradley Leibovich Abhinav KhannaAbhinav Khanna More articles by this author , Vidit SharmaVidit Sharma More articles by this author , Adriana GregoryAdriana Gregory More articles by this author , H. Chase GottlichH. Chase Gottlich More articles by this author , Cole J. CookCole J. Cook More articles by this author , Jason KlugJason Klug More articles by this author , Christine LohseChristine Lohse More articles by this author , Theodora PotretzkeTheodora Potretzke More articles by this author , Aaron PotretzkeAaron Potretzke More articles by this author , Stephen A. BoorjianStephen A. Boorjian More articles by this author , R. Houston ThompsonR. Houston Thompson More articles by this author , Naoki TakahashiNaoki Takahashi More articles by this author , Bradley EricksonBradley Erickson More articles by this author , John ChevilleJohn Cheville More articles by this author , Timothy KlineTimothy Kline More articles by this author , and Bradley LeibovichBradley Leibovich More articles by this author View All Author Informationhttps://doi.org/10.1097/JU.0000000000003239.04AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: Up to 20% of renal tumors are benign and may not require treatment. However, benign versus malignant renal tumors cannot be distinguished using cross-sectional imaging. Renal mass biopsy is a possible solution, but biopsy is invasive and has notable non-diagnostic and false negative rates. As a result, many patients proceed directly to treatment, including some who undergo extirpative surgery for benign tumors. We aim to develop a radiomics and machine learning model for distinguishing between oncocytoma versus malignant renal neoplasms based on abdominal CT images. METHODS: Our institutional registry was queried for patients who underwent surgical treatment of renal tumors from 2000-2018. All surgical specimens underwent pathology re-review by an expert genitourinary pathologist. A total of 843 images from 609 patients (434 images of oncocytoma and 409 images of malignant renal tumor) were included. A previously developed artificial intelligence algorithm for segmentation of kidney, cyst, and tumor area was applied. Images were preprocessed by resampling via linear interpolation to 0.8 mm x 0.8 mm x 5 mm, window/level=440/40, and intensity normalized [0 255]. Features were extracted for the tumor region using PyRadiomics with a fixed bin width of 16. Both unsupervised (PCA, tSNE) and supervised (LR, SVM) machine learning approaches were explored. Data was split 80:20, and cross validation was used in the training/validation set parameters. The best performing model based on F1 macro was then calibrated using an isotonic calibration approach. RESULTS: Dimensionality reduction showed adequate class separation (Figure 1A). The top 10 radiomic features showed a mixture of first and second order features (Figure 1B). The final model reached an accuracy of 0.90 (Figure 1C). Of the 168 images in the testing set, only 2 benign cases were predicted to be malignant, and 15 malignant cases were predicted to be benign. The AUC for oncocytoma versus malignant histology prediction was 0.96 with CI [0.93 0.98] (Figure 1D). CONCLUSIONS: We developed a machine learning model for accurately distinguishing benign oncocytoma versus malignant renal tumors based on CT images alone. This may provide an opportunity for non-invasive risk stratification of solid renal neoplasms, which could potentially reduce overtreatment of renal tumors. Source of Funding: None © 2023 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 209Issue Supplement 4April 2023Page: e233 Advertisement Copyright & Permissions© 2023 by American Urological Association Education and Research, Inc.MetricsAuthor Information Abhinav Khanna More articles by this author Vidit Sharma More articles by this author Adriana Gregory More articles by this author H. Chase Gottlich More articles by this author Cole J. Cook More articles by this author Jason Klug More articles by this author Christine Lohse More articles by this author Theodora Potretzke More articles by this author Aaron Potretzke More articles by this author Stephen A. Boorjian More articles by this author R. Houston Thompson More articles by this author Naoki Takahashi More articles by this author Bradley Erickson More articles by this author John Cheville More articles by this author Timothy Kline More articles by this author Bradley Leibovich More articles by this author Expand All Advertisement PDF downloadLoading ...