Public Release Type:
Journal
Publication Year: 2023
Affiliation: aComputer Assisted Clinical Medicine, Medical Faculty Mannheim, Heidelberg University, Mannheim, 68167, Baden Württemberg, Germany
bMannheim Institute for Intelligent Systems in Medicine, Medical Faculty Mannheim, Heidelberg University, Mannheim, 68167, Baden Württemberg, Germany
cDepartment of Radiology and Nuclear Medicine, University Medical Centre Mannheim, Medical Faculty Mannheim, Heidelberg University, Mannheim, 68167, Baden Württemberg, Germany
dBioengineering Department, Istituto di Ricerche Farmacologiche Mario Negri IRCCS, Ranica (BG), 24020, Italy
Authors:
Tollens F,
Raj A,
Hansen L,
Caroli A,
Nörenberg D,
Zöllnera FG
Request IDs:
22433
Studies:
Consortium for Radiologic Imaging Studies of Polycystic Kidney Disease
We present an automated deep learning approach integrating MRI and conventional clinical markers to predict renal function decline after eight years. We feed MRIs of segmented kidneys to a convolutional neural network. Simultaneously, we use the HtTKV, age, and eGFR at the baseline visit as input to a multi-layer perceptron. Finally, we combine their output and run them through a final MLP to make our prognosis. Results show that our approach could produce a precision/recall at 90% and an AUC above 0.95. Summary of Main Findings: We present an automated deep learning approach combining information from MR imaging and conventional clinical markers to predict renal function in ADPKD decline after eight years with a precision and recall of 90% and an AUC of 0.95.