Meeting Banner
Abstract #4812

Quantifying Domain Shift for Deep Learning Based Synthetic CT Generation with Uncertainty Predictions

Matt Hemsley1, Liam S.P Lawrence1, Rachel W Chan2, Brige Chuge3,4, and Angus Z Lau1,2
1Medical Biophysics, University of Toronto, Toronto, ON, Canada, 2Physical Sciences, Sunnybrook Research Insitute, Toronto, ON, Canada, 3Department of Radiation Oncology, Sunnybrook Health Sciences Centre, Toronto, ON, Canada, 4Department of Physics, Ryerson University, Toronto, ON, Canada


Convolutional Neural Networks behave unpredictively when test images differ from the training images, for example when different sequences or acquisition parameters are used. We trained models to generate synthetic CT images, and tested the models on both in-distribution and out-of-distribution input sequences to determine the magnitude of performance loss. Additionally, we evaluated if uncertainty estimates made using dropout-based variational inference could detect spatial regions of failure. Networks tested on out of distribution images failed to generate accurate synthetic CT images. Uncertainty estimates identified spatial regions of failure and increased with the difference between the training and testing sets.

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

Join Here