Florian Knoll1
1FAU Erlangen Nuremberg, Germany
Synopsis
Keywords: Musculoskeletal: Joints, Image acquisition: Machine learning, Image acquisition: Artefacts
This talk will highlight
potential issues during the translation of Artificial Intelligence (AI) methods
from basic science to clinical use. Using selected examples, we will discuss these
challenges and how they can be addressed.
Highlights
Introduce the
potential that AI methods have for musculoskeletal MRI, while at the same time
pointing out issues that can be encountered.Target audience
Clinicians and
researchers interested in AI methods for musculoskeletal MRI.Outcome/Objective
To provide an
overview of the opportunities and challenges associated with the use of AI for musculoskeletal
MRI.Purpose
Artificial
Intelligence (AI) methods, in particular Deep Learning (DL)1 methods
have shown substantial potential to improve the workflow in all aspects of
medical imaging, from data acquisition to image reconstruction, image
enhancement, image analysis and diagnostic classification2. However,
the translation of these basic science developments to tools that can really be
used in daily clinical practice is not without challenges. DL models require
large amounts of training data and computational resources for training. This
not only leads to substantial costs associated with data curation, labeling and
high-performance computing, but also raises legal questions regarding patient
privacy. There are also methodological challenges that can lead to failure of
DL-models during deployment. These are mostly due to unexpected domain shifts
between training and inference after deployment, inconsistencies in the input
data caused by incorrect data preprocessing or general misuse of the technology.
In this lecture, we will cover examples for these challenges from the areas of
musculoskeletal image acquisition, reconstruction, and analysis. In particular,
we will discuss findings from the recent fastMRI challenges on image
reconstruction for accelerated acquisitions3,4,5. We will also cover
future research directions that are targeted at overcoming these challenges.Acknowledgements
The author
acknowledges funding support from NIH R01EB024532, NIH P41EB017183, DFG
513220538 and DFG 512819079. The author also gratefully acknowledges the
scientific support and HPC resources provided by the Erlangen National High
Performance Computing Center (NHR@FAU) of the Friedrich-Alexander-Universität
Erlangen-Nürnberg (FAU) under the NHR project b143dc. NHR funding is provided
by federal and Bavarian state authorities. NHR@FAU hardware is partially funded
by the German Research Foundation (DFG) – 440719683.References
1. Y. LeCun, Y. Bengio, and G. Hinton, “Deep
learning,” Nature, vol. 521, no. 7553, pp. 436–444, 2015.
2. G. Wang, “Perspective on Deep Imaging”, IEEE
Access 8914-8924 (2016).
3. Knoll F, Zbontar J, Sriram A,
Muckley MJ, Bruno M, Defazio A, Parente M, Geras KJ, Katsnelson J, Chandarana
H, Zhang Z, Drozdzal M, Romero A, Rabbat M, Vincent P, Pinkerton J, Wang D,
Yakubova N, Owens E, Zitnick CL, Recht MP, Sodickson DK, Lui YW. fastMRI: a
publicly available raw k-space and DICOM dataset for accelerated MR image
reconstruction using machine learning. Radiology Artificial Intelligence
(2:2020).
4. Knoll F, Murrel T, Sriram A, Yakubova N,
Zbontar J, Rabbat M, Defazio A, Muckley MJ, Sodickson DK, Zitnick CL, Recht MP.
Advancing machine learning for MR image reconstruction with an open
competition: Overview of the 2019 fastMRI challenge. Magnetic Resonance in
Medicine, 84:3054–3070 (2020).
5. Muckley M,
Riemenschneider B, Radmanesh A, Kim S, Jeong G, Ko J, Jun Y, Shin H, Hwang D,
Mostapha M, Arberet S, Nickel D, Ramzi Z, Ciuciu P, Starck JL, Teuwen J,
Karkalousos D, Zhang C, Sriram A, Huang Z, Yakubova N, Lui Y, Knoll F. Results
of the 2020 fastMRI Challenge for Machine Learning MR Image Reconstruction.
IEEE Transactions on Medical Imaging 40: 2306-2317 (2021).