Jérémy Dana1,2,3,4, Thierry L Lefebvre5,6, Peter Savadjiev4,5,7,8,9, Sylvain Bodard1,10, Simon Gauvin4,11, Sahir Rai Bhatnagar4,12, Reza Forghani4,9,11, Olivier Hélénon1,10, Caroline Reinhold13,14,15. 1. Assistance Publique - Hôpitaux de Paris, Paris University, Paris, France. 2. Inserm U1110, Institut de Recherche Sur Les Maladies Virales Et Hépatiques, Strasbourg University, Strasbourg, France. 3. Institute of Image-Guided Surgery, University Hospital Institute, Strasbourg, France. 4. Department of Diagnostic Radiology, McGill University, Montreal, Canada. 5. Medical Physics Unit, McGill University, Montreal, Canada. 6. Cancer Research UK Cambridge Institute, University of Cambridge, Cambridge, UK. 7. School of Computer Science, McGill University, Montreal, Canada. 8. Department of Pathology, McGill University, Montreal, Canada. 9. Augmented Intelligence & Precision Health Laboratory of the Research Institute of McGill University Health Centre, Montreal, Canada. 10. Department of Adult Radiology, Necker Enfant-Malades University Hospital, Assistance Publique Des Hôpitaux de Paris, Paris, France. 11. Montreal Imaging Experts Inc, Montreal, Canada. 12. Department of Epidemiology, Biostatistics and Occupational Health, McGill University, Montreal, Canada. 13. Department of Diagnostic Radiology, McGill University, Montreal, Canada. caroline.reinhold@mcgill.ca. 14. Augmented Intelligence & Precision Health Laboratory of the Research Institute of McGill University Health Centre, Montreal, Canada. caroline.reinhold@mcgill.ca. 15. Montreal Imaging Experts Inc, Montreal, Canada. caroline.reinhold@mcgill.ca.
Abstract
OBJECTIVE: To distinguish benign from malignant cystic renal lesions (CRL) using a contrast-enhanced CT-based radiomics model and a clinical decision algorithm. METHODS: This dual-center retrospective study included patients over 18 years old with CRL between 2005 and 2018. The reference standard was histopathology or 4-year imaging follow-up. Training and testing datasets were acquired from two institutions. Quantitative 3D radiomics analyses were performed on nephrographic phase CT images. Ten-fold cross-validated LASSO regression was applied to the training dataset to identify the most discriminative features. A logistic regression model was trained to classify malignancy and tested on the independent dataset. Reported metrics included areas under the receiver operating characteristic curves (AUC) and balanced accuracy. Decision curve analysis for stratifying patients for surgery was performed in the testing dataset. A decision algorithm was built by combining consensus radiological readings of Bosniak categories and radiomics-based risks. RESULTS: A total of 149 CRL (139 patients; 65 years [56-72]) were included in the training dataset-35 Bosniak(B)-IIF (8.6% malignancy), 23 B-III (43.5%), and 23 B-IV (87.0%)-and 50 CRL (46 patients; 61 years [51-68]) in the testing dataset-12 B-IIF (8.3%), 10 B-III (60.0%), and 9 B-IV (100%). The machine learning model achieved high diagnostic performance in predicting malignancy in the testing dataset (AUC = 0.96; balanced accuracy = 94%). There was a net benefit across threshold probabilities in using the clinical decision algorithm over management guidelines based on Bosniak categories. CONCLUSION: CT-based radiomics modeling accurately distinguished benign from malignant CRL, outperforming the Bosniak classification. The decision algorithm best stratified lesions for surgery and active surveillance. KEY POINTS: • The radiomics model achieved excellent diagnostic performance in identifying malignant cystic renal lesions in an independent testing dataset (AUC = 0.96). • The machine learning-enhanced decision algorithm outperformed the management guidelines based on the Bosniak classification for stratifying patients to surgical ablation or active surveillance.
OBJECTIVE: To distinguish benign from malignant cystic renal lesions (CRL) using a contrast-enhanced CT-based radiomics model and a clinical decision algorithm. METHODS: This dual-center retrospective study included patients over 18 years old with CRL between 2005 and 2018. The reference standard was histopathology or 4-year imaging follow-up. Training and testing datasets were acquired from two institutions. Quantitative 3D radiomics analyses were performed on nephrographic phase CT images. Ten-fold cross-validated LASSO regression was applied to the training dataset to identify the most discriminative features. A logistic regression model was trained to classify malignancy and tested on the independent dataset. Reported metrics included areas under the receiver operating characteristic curves (AUC) and balanced accuracy. Decision curve analysis for stratifying patients for surgery was performed in the testing dataset. A decision algorithm was built by combining consensus radiological readings of Bosniak categories and radiomics-based risks. RESULTS: A total of 149 CRL (139 patients; 65 years [56-72]) were included in the training dataset-35 Bosniak(B)-IIF (8.6% malignancy), 23 B-III (43.5%), and 23 B-IV (87.0%)-and 50 CRL (46 patients; 61 years [51-68]) in the testing dataset-12 B-IIF (8.3%), 10 B-III (60.0%), and 9 B-IV (100%). The machine learning model achieved high diagnostic performance in predicting malignancy in the testing dataset (AUC = 0.96; balanced accuracy = 94%). There was a net benefit across threshold probabilities in using the clinical decision algorithm over management guidelines based on Bosniak categories. CONCLUSION: CT-based radiomics modeling accurately distinguished benign from malignant CRL, outperforming the Bosniak classification. The decision algorithm best stratified lesions for surgery and active surveillance. KEY POINTS: • The radiomics model achieved excellent diagnostic performance in identifying malignant cystic renal lesions in an independent testing dataset (AUC = 0.96). • The machine learning-enhanced decision algorithm outperformed the management guidelines based on the Bosniak classification for stratifying patients to surgical ablation or active surveillance.
Authors: Hendrik Van Poppel; Luigi Da Pozzo; Walter Albrecht; Vsevolod Matveev; Aldo Bono; Andrzej Borkowski; Jean-Marie Marechal; Laurence Klotz; Eila Skinner; Thomas Keane; Ilse Claessens; Richard Sylvester Journal: Eur Urol Date: 2006-11-15 Impact factor: 20.096
Authors: Maxine Sun; Marco Bianchi; Jens Hansen; Quoc-Dien Trinh; Firas Abdollah; Zhe Tian; Jesse Sammon; Shahrokh F Shariat; Markus Graefen; Francesco Montorsi; Paul Perrotte; Pierre I Karakiewicz Journal: Eur Urol Date: 2012-03-31 Impact factor: 20.096
Authors: Stuart G Silverman; Ivan Pedrosa; James H Ellis; Nicole M Hindman; Nicola Schieda; Andrew D Smith; Erick M Remer; Atul B Shinagare; Nicole E Curci; Steven S Raman; Shane A Wells; Samuel D Kaffenberger; Zhen J Wang; Hersh Chandarana; Matthew S Davenport Journal: Radiology Date: 2019-06-18 Impact factor: 11.105
Authors: Wooil Kim; Sang Min Lee; Jung Im Kim; Yura Ahn; Sohee Park; Jooae Choe; Joon Beom Seo Journal: AJR Am J Roentgenol Date: 2021-10-20 Impact factor: 3.959
Authors: Sabina Sevcenco; Claudio Spick; Thomas H Helbich; Gertraud Heinz; Shahrokh F Shariat; Hans C Klingler; Michael Rauchenwald; Pascal A Baltzer Journal: Eur Radiol Date: 2016-10-19 Impact factor: 5.315