Play all

Intro

Magnetic Resonance Imaging

Data Pipeline

Raw Data in MRI

k-space Sampling

Constraints

Can We Speed Things Up??

MRI Super-Resolution Convert low resolution (LR) to high resolution (HR)

Domain Knowledge • Embedding physical principles into the model

Reducing Data Requirements Domain knowledge reduce extent of ill-posed inverse problems

What if we have no training data

Deep Image Prior

Convolutional Decoder ConvDe

Semi-Supervised Learning: Self-Tr

Self-Training for MRI Recon

Why is ConvDecoder Better? Zero Unrolled ConvDecoder

ConvDecoder Noisy Student

Self-Training Takeaway Keep forward model identical; modify pseudo-labels

Invariance to Forward Model Change Supervised Unsupervised

Generalized Consistency Framew

Artifact Invariant Reconstruction

Metrics for Evaluation Discordance between quantitative and qualitative metrics

New Multi-Task Dataset

Surrogates without Dense Labels Use self-supervised tasks to build image representations

Self-Supervised Image Quality

Overall Takeaway • Adding domain knowledges reduces data requirements

Open Source Data/Code

Conclusions

Description:

Explore a comprehensive lecture on data-efficient AI techniques for accelerating MRI acquisition. Delve into the challenges of medical image acquisition and learn how physics-guided AI is revolutionizing the speed of magnetic resonance imaging. Discover novel unsupervised and semi-supervised approaches that reduce the need for extensive paired datasets in supervised model training. Gain insights into a newly released 1.5TB dataset for evaluating MRI reconstructions using clinically-relevant metrics. Examine topics such as MRI super-resolution, domain knowledge integration, deep image prior, self-training methods, and artifact-invariant reconstruction. Understand the importance of evaluation metrics and explore a new multi-task dataset for comprehensive assessment. Learn about self-supervised tasks for building image representations and their application in image quality assessment. Grasp the significance of incorporating domain knowledge to reduce data requirements in AI-driven MRI acceleration techniques. Read more

Data-Efficient AI for Accelerating MRI Acquisition

Stanford University

Add to list

#Computer Science #Artificial Intelligence #Machine Learning #Unsupervised Learning #Semi-supervised Learning

0:00 / 0:00