A Decoupled Uncertainty Model for MRI Segmentation Quality Estimation

Quality control (QC) of MR images is essential to ensure that downstream analyses such as segmentation can be performed successfully. Currently, QC is predominantly performed visually and subjectively, at significant time and operator cost. We aim to automate the process using a probabilistic network that estimates segmentation uncertainty through a heteroscedastic noise model, providing a measure of task-specific quality. By augmenting training images with k-space artefacts, we propose a novel CNN architecture to decouple sources of uncertainty related to the task and different k-space artefacts in a self-supervised manner. This enables the prediction of separate uncertainties for different types of data degradation. While the uncertainty predictions reflect the presence and severity of artefacts, the network provides more robust and generalisable segmentation predictions given the quality of the data. We show that models trained with artefact augmentation provide informative measures of uncertainty on both simulated artefacts and problematic realworld images identified by human-raters, both qualitatively and quantitatively in the form of error bars on volume measurements. Relating artefact uncertainty to segmentation Dice scores, we observe that our uncertainty predictions provide a better estimate of MRI quality from the point of view of the task (gray matter segmentation) compared to commonly used metrics of quality including signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR), hence providing a real-time quality metric indicative of segmentation quality.

[1]  Irene Brusini,et al.  A deep learning-based pipeline for error detection and quality control of brain MRI segmentation results , 2020 .

[2]  Anant Madabhushi,et al.  MRQy: An Open-Source Tool for Quality Control of MR Imaging Data , 2020, ArXiv.

[3]  Quoc V. Le,et al.  RandAugment: Practical data augmentation with no separate search , 2019, ArXiv.

[4]  Quoc V. Le,et al.  Self-Training With Noisy Student Improves ImageNet Classification , 2019, 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[5]  Parashkev Nachev,et al.  Computer Methods and Programs in Biomedicine NiftyNet: a deep-learning platform for medical imaging , 2022 .

[6]  Jason Dowling,et al.  An open source automatic quality assurance (OSAQA) tool for the ACR MRI phantom , 2014, Australasian Physical & Engineering Sciences in Medicine.

[7]  Balaji Lakshminarayanan,et al.  AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty , 2020, ICLR.

[8]  H. Rolf Jäger,et al.  Let's agree to disagree: learning highly debatable multirater labelling , 2019, MICCAI.

[9]  Daniel Rueckert,et al.  Geodesic Information Flows: Spatially-Variant Graphs and Their Application to Segmentation and Fusion , 2015, IEEE Transactions on Medical Imaging.

[10]  Jan Kautz,et al.  Loss Functions for Image Restoration With Neural Networks , 2017, IEEE Transactions on Computational Imaging.

[11]  Antonio Criminisi,et al.  Uncertainty Quantification in Deep Learning for Safer Neuroimage Enhancement , 2019, ArXiv.

[12]  M. Jorge Cardoso,et al.  As easy as 1, 2... 4? Uncertainty in counting tasks for medical imaging , 2019, MICCAI.

[13]  Alexei A. Efros,et al.  Image-to-Image Translation with Conditional Adversarial Networks , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[14]  Henkjan Huisman,et al.  Supervised Uncertainty Quantification for Segmentation with Multiple Annotations , 2019, MICCAI.

[15]  Harri Valpola,et al.  Weight-averaged consistency targets improve semi-supervised deep learning results , 2017, ArXiv.

[16]  Seong Joon Oh,et al.  CutMix: Regularization Strategy to Train Strong Classifiers With Localizable Features , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).

[17]  Nassir Navab,et al.  Bayesian QuickNAT: Model uncertainty in deep whole-brain segmentation for structure-wise quality control , 2018, NeuroImage.

[18]  Hui Zhang,et al.  A supervised learning approach for diffusion MRI quality control with minimal training data , 2018, NeuroImage.

[19]  Sébastien Ourselin,et al.  Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neural networks , 2018, Neurocomputing.

[20]  Graham W. Taylor,et al.  Leveraging Uncertainty Estimates for Predicting Segmentation Quality , 2018, ArXiv.

[21]  Krzysztof J. Gorgolewski,et al.  MRIQC: Advancing the automatic prediction of image quality in MRI from unseen sites , 2016, bioRxiv.

[22]  Zoubin Ghahramani,et al.  Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning , 2015, ICML.

[23]  Douglas L Arnold,et al.  Automated quality control of brain MR images , 2008, Journal of magnetic resonance imaging : JMRI.

[24]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[25]  Jiachen Zhuo,et al.  AAPM/RSNA physics tutorial for residents: MR artifacts, safety, and quality control. , 2006, Radiographics : a review publication of the Radiological Society of North America, Inc.

[26]  Roberto Cipolla,et al.  Multi-task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semantics , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[27]  M. Jorge Cardoso,et al.  Uncertainty in multitask learning: joint representations for probabilistic MR-only radiotherapy planning , 2018, MICCAI.

[28]  Geoffrey E. Hinton,et al.  Distilling the Knowledge in a Neural Network , 2015, ArXiv.

[29]  Sebastien Ourselin,et al.  A k-Space Model of Movement Artefacts: Application to Segmentation Augmentation and Artefact Removal , 2020, IEEE Transactions on Medical Imaging.

[30]  Klaus H. Maier-Hein,et al.  Abstract: nnU-Net: Self-adapting Framework for U-Net-Based Medical Image Segmentation , 2019, Bildverarbeitung für die Medizin.

[31]  M. Jorge Cardoso,et al.  A Heteroscedastic Uncertainty Model for Decoupling Sources of MRI Image Quality , 2020, MIDL.

[32]  Klaus H. Maier-Hein,et al.  A Probabilistic U-Net for Segmentation of Ambiguous Images , 2018, NeurIPS.

[33]  Marc Combalia,et al.  Uncertainty Estimation in Deep Neural Networks for Dermoscopic Image Classification , 2020, 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW).

[34]  Charles Blundell,et al.  Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles , 2016, NIPS.

[35]  Makarand V. Ratnaparkhi,et al.  On the functional relationship between entropy and variance with related applications , 1986 .

[36]  Daniel Rueckert,et al.  Deep Generative Model-based Quality Control for Cardiac MRI Segmentation , 2020, MICCAI.

[37]  Alex Kendall,et al.  What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision? , 2017, NIPS.

[38]  Lalitha Venkataramanan,et al.  Dual Neural Network Architecture for Determining Epistemic and Aleatoric Uncertainties , 2019, ArXiv.