Bayesian image quality transfer with CNNs: Exploring uncertainty in dMRI super-resolution
In this work, we investigate the value of uncertainty modeling in 3D super-resolution with convolutional neural networks (CNNs). Deep learning has shown success in a plethora of medical image trans- formation problems, such as super-resolution (SR) and image synthesis. However, the highly ill-posed...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Conference item |
| Published: |
Springer
2017
|
| _version_ | 1797089998717059072 |
|---|---|
| author | Tanno, R Worrall, D Ghosh, A Kaden, E Sotiropoulos, S Criminisi, A Alexander, D |
| author_facet | Tanno, R Worrall, D Ghosh, A Kaden, E Sotiropoulos, S Criminisi, A Alexander, D |
| author_sort | Tanno, R |
| collection | OXFORD |
| description | In this work, we investigate the value of uncertainty modeling in 3D super-resolution with convolutional neural networks (CNNs). Deep learning has shown success in a plethora of medical image trans- formation problems, such as super-resolution (SR) and image synthesis. However, the highly ill-posed nature of such problems results in inevitable ambiguity in the learning of networks. We propose to account for intrinsic uncertainty through a per-patch heteroscedastic noise model and for parameter uncertainty through approximate Bayesian inference in the form of variational dropout. We show that the combined benefits of both lead to the state-of-the-art performance SR of diffusion MR brain images in terms of errors compared to ground truth. We further show that the reduced error scores produce tangible benefits in downstream tractography. In addition, the probabilistic nature of the methods naturally confers a mechanism to quantify uncertainty over the super-resolved output. We demonstrate through experiments on both healthy and pathological brains the potential utility of such an uncertainty measure in the risk assessment of the super-resolved images for subsequent clinical use. |
| first_indexed | 2024-03-07T03:12:08Z |
| format | Conference item |
| id | oxford-uuid:b48b7fc2-85cf-4496-be39-0e72540b63bb |
| institution | University of Oxford |
| last_indexed | 2024-03-07T03:12:08Z |
| publishDate | 2017 |
| publisher | Springer |
| record_format | dspace |
| spelling | oxford-uuid:b48b7fc2-85cf-4496-be39-0e72540b63bb2022-03-27T04:26:57ZBayesian image quality transfer with CNNs: Exploring uncertainty in dMRI super-resolutionConference itemhttp://purl.org/coar/resource_type/c_5794uuid:b48b7fc2-85cf-4496-be39-0e72540b63bbSymplectic Elements at OxfordSpringer2017Tanno, RWorrall, DGhosh, AKaden, ESotiropoulos, SCriminisi, AAlexander, DIn this work, we investigate the value of uncertainty modeling in 3D super-resolution with convolutional neural networks (CNNs). Deep learning has shown success in a plethora of medical image trans- formation problems, such as super-resolution (SR) and image synthesis. However, the highly ill-posed nature of such problems results in inevitable ambiguity in the learning of networks. We propose to account for intrinsic uncertainty through a per-patch heteroscedastic noise model and for parameter uncertainty through approximate Bayesian inference in the form of variational dropout. We show that the combined benefits of both lead to the state-of-the-art performance SR of diffusion MR brain images in terms of errors compared to ground truth. We further show that the reduced error scores produce tangible benefits in downstream tractography. In addition, the probabilistic nature of the methods naturally confers a mechanism to quantify uncertainty over the super-resolved output. We demonstrate through experiments on both healthy and pathological brains the potential utility of such an uncertainty measure in the risk assessment of the super-resolved images for subsequent clinical use. |
| spellingShingle | Tanno, R Worrall, D Ghosh, A Kaden, E Sotiropoulos, S Criminisi, A Alexander, D Bayesian image quality transfer with CNNs: Exploring uncertainty in dMRI super-resolution |
| title | Bayesian image quality transfer with CNNs: Exploring uncertainty in dMRI super-resolution |
| title_full | Bayesian image quality transfer with CNNs: Exploring uncertainty in dMRI super-resolution |
| title_fullStr | Bayesian image quality transfer with CNNs: Exploring uncertainty in dMRI super-resolution |
| title_full_unstemmed | Bayesian image quality transfer with CNNs: Exploring uncertainty in dMRI super-resolution |
| title_short | Bayesian image quality transfer with CNNs: Exploring uncertainty in dMRI super-resolution |
| title_sort | bayesian image quality transfer with cnns exploring uncertainty in dmri super resolution |
| work_keys_str_mv | AT tannor bayesianimagequalitytransferwithcnnsexploringuncertaintyindmrisuperresolution AT worralld bayesianimagequalitytransferwithcnnsexploringuncertaintyindmrisuperresolution AT ghosha bayesianimagequalitytransferwithcnnsexploringuncertaintyindmrisuperresolution AT kadene bayesianimagequalitytransferwithcnnsexploringuncertaintyindmrisuperresolution AT sotiropouloss bayesianimagequalitytransferwithcnnsexploringuncertaintyindmrisuperresolution AT criminisia bayesianimagequalitytransferwithcnnsexploringuncertaintyindmrisuperresolution AT alexanderd bayesianimagequalitytransferwithcnnsexploringuncertaintyindmrisuperresolution |