Synopsis

Keywords: Image acquisition: Machine learning, Image acquisition: Image processing, Image acquisition: Reconstruction

Artificial intelligence/machine learning (AI/ML) based techniques have gathered interest as a possible means to improve MRI processing pipelines, with applications ranging from image reconstruction from raw data to extraction of quantitative biomarkers from imaging data in post-processing. Our purpose is to review existing and emerging AI/ML methods for various MRI processing applications.

Target Audience

Physicians and engineers interested in developing and translating emerging AI/ML methods into their processing pipelines for computational MRI and data analysis, and in understanding the nuances and implications of using such methods.

Objectives

Understand the emerging AI/ML methods and their application to processing of MRI datasets.

Syllabus

Artificial intelligence/machine learning (AI/ML) based techniques have gathered interest as a possible means to improve MRI processing pipelines, with applications ranging from image reconstruction from raw data to extraction of quantitative biomarkers from imaging data in post-processing. Our purpose is to review existing and emerging AI/ML methods for various MRI processing applications.
In this lecture, we will overview the technical underpinnings of well-established ideas in convolutional neural networks and supervised training, as well as newer approaches related to attention mechanisms, transformers, diffusion models and unsupervised learning strategies. We will also present the application of these methods in various applications. For computational MRI, we will provide background in physics-driven AI/ML methods for image reconstruction, as well as image denoising, and discuss the state-of-the-art methods for these tasks. For post-processing applications, we will overview image segmentation, registration and analysis tasks, and review recent techniques for these problems.

Takeaway Messages

Following this lecture, participants should be able to:
• Understand the technical underpinnings of emerging AI/ML models and new learning strategies
• Understand the application of these techniques to a variety of MRI processing problems, ranging from processing of raw data to image analysis

Acknowledgements

The author’s work is supported by NIH R01HL153146, NIH R01EB032830, NIH R21EB028369, NIH P41EB027061, NIH U01EB025144, and NSF CAREER CCF-1651825.

References

No reference found.