Meeting Banner
Abstract #0476

Detection of kidney cysts using a Convolutional Neural Network on 11,000 MRI data sets from the German National Cohort

Wilfried Reichardt1, Jan Lipovsek1, Marco Reisert1, Harald Horbach1, Christopher Schlett1, Fabian Bamberg1, Peggy Sekula1, Anna Köttgen1, Elias Kellner1, and Martin Büchert1
1University Medical Center Freiburg, Freiburg, Germany


We implemented a processing pipeline for kidney cyst segmentation using a hierarchical patch-based stack of U-nets and applied it to abdominal MRI images of the German National Cohort (GNC) study. The training data set included 300 cases, and the final net was applied to the dataset of 11,207 MRIs. Kidney cysts could be segmented with 98% sensitivity above a lower detection threshold of 1ml. The relation of first parameters based on the cyst segmentation with age and sex are presented. This result presents an optimal starting point to identify more advanced bimarkaers and their correlates, especially with kidney function parameters.

This abstract and the presentation materials are available to members only; a login is required.

Join Here