Towards lower-dose PET using physics-based uncertainty-aware multimodal learning with robustness to out-of-distribution data

Sudarshan, Viswanath P.; Upadhyay, Uddeshya; Egan, Gary F.; Chen, Zhaolin; Awate, Suyash P.

doi:10.1016/j.media.2021.102187

Cited by 30 publications

(15 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A comprehensive standard for the evaluation of deep learning-based PET image processing algorithms is needed to provide a quantitative and comparable evaluation of the performance of models. The standard should include an evaluation of the model generalisability under varying conditions, as was shown in some of the studies reviewed in this work [ 89 , 102 ].…”

Section: Discussionmentioning

confidence: 99%

“…Sudarshan et al [ 102 ] trained a modified Unet to map × 180 low-dose 18 F-FDG PET brain images with coregistered T 1 and T 2 MRI to full-dose PET images and uncertainty maps using an uncertainty aware loss functions in both image space and sinogram space. Training the uncertainty estimator using a Bayesian framework did not require ground truth uncertainty maps.…”

Section: Review Of Deep Learning-based Low-dose To Full-dose Post-pro...mentioning

confidence: 99%

See 1 more Smart Citation

Deep learning-based image reconstruction and post-processing methods in positron emission tomography for low-dose imaging and resolution enhancement

Pain

Egan

Chen

2022

Eur J Nucl Med Mol Imaging

Self Cite

View full text Add to dashboard Cite

Image processing plays a crucial role in maximising diagnostic quality of positron emission tomography (PET) images. Recently, deep learning methods developed across many fields have shown tremendous potential when applied to medical image enhancement, resulting in a rich and rapidly advancing literature surrounding this subject. This review encapsulates methods for integrating deep learning into PET image reconstruction and post-processing for low-dose imaging and resolution enhancement. A brief introduction to conventional image processing techniques in PET is firstly presented. We then review methods which integrate deep learning into the image reconstruction framework as either deep learning-based regularisation or as a fully data-driven mapping from measured signal to images. Deep learning-based post-processing methods for low-dose imaging, temporal resolution enhancement and spatial resolution enhancement are also reviewed. Finally, the challenges associated with applying deep learning to enhance PET images in the clinical setting are discussed and future research directions to address these challenges are presented.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Review Of Deep Learning-based Low-dose To Full-dose Post-pro...mentioning

confidence: 99%

Deep learning-based image reconstruction and post-processing methods in positron emission tomography for low-dose imaging and resolution enhancement

Pain

Egan

Chen

2022

Eur J Nucl Med Mol Imaging

Self Cite

View full text Add to dashboard Cite

show abstract

“…There is no reliable definition and standard to define what is "modal" data and the corresponding properties of the data [7].…”

Section: Modality and Storage In Medical Datamentioning

confidence: 99%

Multimodal medical data statistical system based on machine learning algorithm

Zeng

2023

Second International Conference on Biomedical and Intelligent Systems (IC-BIS 2023)

View full text Add to dashboard Cite

“…Tanno and colleagues investigated uncertainty modelling for diffusion MRI super-resolution and sought to provide a high-level explanation of deep learning models with respect to variation in input datasets [ 187 ]. A number of similar works applied explicit uncertainty models during model training and inference in order to assess model robustness and uncertainties associated with input data [ 46 , 188 , 189 ]. Hallucinations are false image features introduced when an imperfect or inaccurate model prior is used during image processing and typically occur when training and testing have different data distributions.…”

Section: Challenges and Future Perspectivesmentioning

confidence: 99%

Deep Learning for Image Enhancement and Correction in Magnetic Resonance Imaging—State-of-the-Art and Challenges

et al. 2022

Self Cite

View full text Add to dashboard Cite

Magnetic resonance imaging (MRI) provides excellent soft-tissue contrast for clinical diagnoses and research which underpin many recent breakthroughs in medicine and biology. The post-processing of reconstructed MR images is often automated for incorporation into MRI scanners by the manufacturers and increasingly plays a critical role in the final image quality for clinical reporting and interpretation. For image enhancement and correction, the post-processing steps include noise reduction, image artefact correction, and image resolution improvements. With the recent success of deep learning in many research fields, there is great potential to apply deep learning for MR image enhancement, and recent publications have demonstrated promising results. Motivated by the rapidly growing literature in this area, in this review paper, we provide a comprehensive overview of deep learning-based methods for post-processing MR images to enhance image quality and correct image artefacts. We aim to provide researchers in MRI or other research fields, including computer vision and image processing, a literature survey of deep learning approaches for MR image enhancement. We discuss the current limitations of the application of artificial intelligence in MRI and highlight possible directions for future developments. In the era of deep learning, we highlight the importance of a critical appraisal of the explanatory information provided and the generalizability of deep learning algorithms in medical imaging.

show abstract

Towards lower-dose PET using physics-based uncertainty-aware multimodal learning with robustness to out-of-distribution data

Cited by 30 publications

References 36 publications

Deep learning-based image reconstruction and post-processing methods in positron emission tomography for low-dose imaging and resolution enhancement

Deep learning-based image reconstruction and post-processing methods in positron emission tomography for low-dose imaging and resolution enhancement

Multimodal medical data statistical system based on machine learning algorithm

Deep Learning for Image Enhancement and Correction in Magnetic Resonance Imaging—State-of-the-Art and Challenges

Contact Info

Product

Resources

About