Mp07-11 Advancements in Prostate Diagnostics: A Composite Method Using 3d Mri and Clinical Data Integration

You have accessJournal of UrologySurgical Technology & Simulation: Artificial Intelligence I (MP07)1 May 2024MP07-11 ADVANCEMENTS IN PROSTATE DIAGNOSTICS: A COMPOSITE METHOD USING 3D MRI AND CLINICAL DATA INTEGRATION Lucas Engelage, Oleksii Bashkanov, Niklas Behnel, Agron Lumiani, Alexander Reich, Paul Ehrlich, Marko Rak, Christian Hansen, Leonhard Steinmeister, and Rolf Muschter Lucas EngelageLucas Engelage , Oleksii BashkanovOleksii Bashkanov , Niklas BehnelNiklas Behnel , Agron LumianiAgron Lumiani , Alexander ReichAlexander Reich , Paul EhrlichPaul Ehrlich , Marko RakMarko Rak , Christian HansenChristian Hansen , Leonhard SteinmeisterLeonhard Steinmeister , and Rolf MuschterRolf Muschter View All Author Informationhttps://doi.org/10.1097/01.JU.0001008728.41882.d7.11AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: The improved diagnostic abilities of advanced computer-assisted imaging, particularly 3D MRI, has significantly improved the identification of prostate abnormalities and prostate cancer diagnosis. However, the diagnostic specificity of other markers, such as prostate-specific antigen (PSA), has been underemphasized due to an over-reliance on radiological assessments, leading to the oversight of critical patient data. Our study aims to bridge this gap by exploring the combined use of imaging and diverse patient data, including demographics and clinical findings, to enhance diagnostic precision. METHODS: This research utilizes a dataset comprising of 3800 biparametric MRI scans from biopsy-naïve patients at ALTA Klinik Bielefeld and accompanying tabular data including birth year, age, weight, height, body mass index and all available PSA levels. We applied fusion techniques to integrate image with demographic and clinical data effectively. The models' predictive capabilities, both binary and multiclass, were evaluated using ROC-Analysis and the quadratic weighted Kappa (QWK). RESULTS: Integrating MR imaging with tabular data, our model demonstrated a ROC-AUC of 0.794 and a QWK of 0.464 in multiclass prognostication of prostatitis and Gleason scores, surpassing the image-only method with a ROC-AUC of 0.736 and a QWK of 0.342. The inclusion of tabular data, particularly three PSA measurements, achieved a ROC of 0.704 and a QWK of 0.339. CONCLUSIONS: A model incorporating MR imaging with clinical and demographic data provided the highest diagnostic accuracy and predictive precision. Our findings advocate for a more integrated approach in computer-aided diagnostic systems, particularly highlighting the value of longitudinal PSA data. This integrative method could significantly improve diagnostic precision for prostate cancer. Source of Funding: This work has been supported, in parts, by the federal state of Saxony-Anhalt (Germany) within the framework of the postgraduates funding and by Central Innovation Programmefor small and medium-sized enterprises (SMEs) Germany © 2024 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 211Issue 5SMay 2024Page: e108 Advertisement Copyright & Permissions© 2024 by American Urological Association Education and Research, Inc.Metrics Author Information Lucas Engelage More articles by this author Oleksii Bashkanov More articles by this author Niklas Behnel More articles by this author Agron Lumiani More articles by this author Alexander Reich More articles by this author Paul Ehrlich More articles by this author Marko Rak More articles by this author Christian Hansen More articles by this author Leonhard Steinmeister More articles by this author Rolf Muschter More articles by this author Expand All Advertisement PDF downloadLoading ...