Abstract #4035

Cascading Classifiers Improve Prostate Segmentation

Ronald James Nowling¹, John Bukowy², Sean D McGarry³, Andrew S Nencka², Jay Urbain^1,4, Allison Lowman², Alexandar Barrington⁵, Mark Hohenwalter², Anjishnu Banerjee⁶, Kenneth A Iczkowski⁷, and Peter S LaViolette^2,5

¹Electrical Engineering and Computer Science, Milwaukee School of Engineering, Milwaukee, WI, United States, ²Radiology, Medical College of Wisconsin, Milwaukee, WI, United States, ³Biophysics, Medical College of Wisconsin, Milwaukee, WI, United States, ⁴Clinical and Translational Sciences Institute, Medical College of Wisconsin, Milwaukee, WI, United States, ⁵Biomedical Engineering, Medical College of Wisconsin, Milwaukee, WI, United States, ⁶Biostatistics, Medical College of Wisconsin, Milwaukee, WI, United States, ⁷Pathology, Medical College of Wisconsin, Milwaukee, WI, United States

We evaluated the U-Net segmentation model on prostate segmentation using data from 39 patients, achieving a Dice score of 73.9%. We improved segmentation performance by applying a convolutional neural network (CNN) to determine whether slices have prostates. Images with prostates are then forwarded to a U-Net model for segmentation. Our two-phase approach achieves a higher Dice score of 85.2%.

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

Join Here