Log In Sign Up

Integrating uncertainty in deep neural networks for MRI based stroke analysis

by   Lisa Herzog, et al.

At present, the majority of the proposed Deep Learning (DL) methods provide point predictions without quantifying the models uncertainty. However, a quantification of the reliability of automated image analysis is essential, in particular in medicine when physicians rely on the results for making critical treatment decisions. In this work, we provide an entire framework to diagnose ischemic stroke patients incorporating Bayesian uncertainty into the analysis procedure. We present a Bayesian Convolutional Neural Network (CNN) yielding a probability for a stroke lesion on 2D Magnetic Resonance (MR) images with corresponding uncertainty information about the reliability of the prediction. For patient-level diagnoses, different aggregation methods are proposed and evaluated, which combine the single image-level predictions. Those methods take advantage of the uncertainty in image predictions and report model uncertainty at the patient-level. In a cohort of 511 patients, our Bayesian CNN achieved an accuracy of 95.33 2 95.89 in aggregation models resulted in higher uncertainty measures to false patient classifications, which enabled to filter critical patient diagnoses that are supposed to be closer examined by a medical doctor. We therefore recommend using Bayesian approaches not only for improved image-level prediction and uncertainty estimation but also for the detection of uncertain aggregations at the patient-level.


page 3

page 4

page 8


Analyzing the Role of Model Uncertainty for Electronic Health Records

In medicine, both ethical and monetary costs of incorrect predictions ca...

Geometric Uncertainty in Patient-Specific Cardiovascular Modeling with Convolutional Dropout Networks

We propose a novel approach to generate samples from the conditional dis...

Towards Data Assimilation in Level-Set Wildfire Models Using Bayesian Filtering

The level-set method is a prominent approach to modelling the evolution ...