Uncertainty Quantification in Deep Learning for Safer Neuroimage Enhancement

Tanno, Ryutaro; Worrall, Daniel; Kaden, Enrico; Ghosh, Aurobrata; Grussu, Francesco; Bizzi, Alberto; Sotiropoulos, Stamatios N.; Criminisi, Antonio; Alexander, Daniel C.

doi:10.48550/arxiv.1907.13418

Cited by 7 publications

(11 citation statements)

References 100 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…ifications to a standard DNN. As shown in prior work (e.g., [3,5]), the final experiment demonstrates how having an estimate of uncertainty adds insight into a quantitative result (see Fig. 4).…”

Section: Discussionsupporting

confidence: 65%

“…Furthermore, capturing one type of uncertainty but not the other is insufficient to estimate the predictive uncertainty-an encompassing measure which describes how well any voxel can be predicted. Uncertainty estimation in image translation, segmentation, and super-resolution has been explored [3,4,5]; however, in this work, we verify that the epistemic and aleatoric uncertainty estimates captured in an image translation task align with the definitions of the terms.…”

Section: Introductionmentioning

confidence: 58%

See 1 more Smart Citation

Validating Uncertainty in Medical Image Translation

Reinhold

Han

et al. 2020

2020 IEEE 17th International Symposium on Biomedical Imaging (ISBI)

View full text Add to dashboard Cite

Medical images are increasingly used as input to deep neural networks to produce quantitative values that aid researchers and clinicians. However, standard deep neural networks do not provide a reliable measure of uncertainty in those quantitative values. Recent work has shown that using dropout during training and testing can provide estimates of uncertainty. In this work, we investigate using dropout to estimate epistemic and aleatoric uncertainty in a CT-to-MR image translation task. We show that both types of uncertainty are captured, as defined, providing confidence in the output uncertainty estimates.

show abstract

Section: Discussionsupporting

confidence: 65%

Section: Introductionmentioning

confidence: 58%

Validating Uncertainty in Medical Image Translation

Reinhold

Han

et al. 2020

2020 IEEE 17th International Symposium on Biomedical Imaging (ISBI)

View full text Add to dashboard Cite

show abstract

“…by means of ensembling of networks [42] or variational dropout [43]. In addition, previous work by Kendall and Gal [44], Tanno et al [45] has shown that the quality of uncertainty estimates can be improved if model (epistemic) and data (aleatoric) uncertainty are assessed simultaneously with separate measures. The current study focused on the assessment of model uncertainty by means of MC-dropout and entropy which is a combination of epistemic and aleatoric uncertainty.…”

Section: Discussionmentioning

confidence: 99%

Automatic segmentation with detection of local segmentation failures in cardiac MRI

Vos¹,

Išgum²

2020

Preprint

View full text Add to dashboard Cite

Segmentation of cardiac anatomical structures in cardiac magnetic resonance images (CMRI) is a prerequisite for automatic diagnosis and prognosis of cardiovascular diseases. To increase robustness and performance of segmentation methods this study combines automatic segmentation and assessment of segmentation uncertainty in CMRI to detect image regions containing local segmentation failures. Three state-of-the-art convolutional neural networks (CNN) were trained to automatically segment cardiac anatomical structures and obtain two measures of predictive uncertainty: entropy and a measure derived by MC-dropout. Thereafter, using the uncertainties another CNN was trained to detect local segmentation failures that potentially need correction by an expert. Finally, manual correction of the detected regions was simulated. Using publicly available CMR scans from the MICCAI 2017 ACDC challenge, the impact of CNN architecture and loss function for segmentation, and the uncertainty measure was investigated. Performance was evaluated using the Dice coefficient and 3D Hausdorff distance between manual and automatic segmentation. The experiments reveal that combining automatic segmentation with simulated manual correction of detected segmentation failures leads to statistically significant performance increase.

show abstract

“…In previous studies, the pixel-wise aleatoric uncertainty can be generated by incorporating Gaussian negative log likelihood (NLL) loss into neural network [28], and applied in MR super-resolution reconstruction [26] [27]. Such uncertainty only represents the uncertainty from data which can not be prevented and it is not the main issue when applying the deep learning based MRI restoration in the clinic practices.…”

Section: E Uncertaintymentioning

confidence: 99%

“…At last, doctors usually concern about the accuracy of restored high-quality images. Tanno et al [26] and Qin et al [27] employed a method to generate aleatoric uncertainty [29] as auxiliary information for doctor to understand the uncertainty of restored MR image. However, such method can not tell if the uncertainty is caused by the noise in the training data or caused by error generated by the deep neural network due to out of distribution (OOD) data, i.e., distribution of training data is not identical to the distribution of test data which is commonly seen in the real clinical environment.…”

Section: Introductionmentioning

confidence: 99%

3D High-Quality Magnetic Resonance Image Restoration in Clinics Using Deep Learning

Li¹,

Liu²

2021

Preprint

View full text Add to dashboard Cite

Shortening acquisition time and reducing the motion-artifact are two of the most essential concerns in magnetic resonance imaging. As a promising solution, deep learning-based high-quality MR image restoration has been investigated to generate higher resolution MR images from lower resolution images acquired with shortened acquisition time and free of motion-artifact, without costing additional acquisition time, modifying the pulse sequences or repeating the acquisition. However, there are still numerous problems preventing deep learning approaches from becoming practical in the clinic environment. Specifically, most of the prior works focus solely on the network model but ignore the impact of various downsampling strategies on the acquisition time. Besides, the long inference time and high GPU consumption are also the bottle neck to deploy most of the prior works in clinics. Furthermore, prior studies employ random movement in retrospective motion artifact generation, resulting in uncontrollable severity of motion artifact. More importantly, doctors are unsure whether the generated MR images are trustworthy, making diagnosis difficult. To overcome all the aforementioned problems, we employed a unified deep learning neural network with 2D convolutional filter for both 3D MRI super resolution and motion artifact reduction, demonstrating such a unified framework can achieve better performance in 3D MRI restoration task compared to other states of the art methods and remains the GPU consumption and inference time significantly low, thus easier to be deployed.We also analyzed several downsampling strategies based on acceleration factor, including multiple combinations of in-plane and through-plane downsampling, and developed a controllable and quantifiable motion artifact generation method. At last, the pixel-wise uncertainty was calculated and used to estimate the accuracy of generated images, providing additional information for a reliable diagnosis.

show abstract

Uncertainty Quantification in Deep Learning for Safer Neuroimage Enhancement

Cited by 7 publications

References 100 publications

Validating Uncertainty in Medical Image Translation

Validating Uncertainty in Medical Image Translation

Automatic segmentation with detection of local segmentation failures in cardiac MRI

3D High-Quality Magnetic Resonance Image Restoration in Clinics Using Deep Learning

Contact Info

Product

Resources

About