Image space formalism of convolutional neural networks for k-space interpolation

2024-02-27

AI Summary

Nonlinear activation functions in k-space interpolation methods (e.g., RAKI) critically influence noise propagation during MRI reconstruction, yet their underlying mechanisms remain poorly understood. Method: We formulate RAKI in the image domain, modeling k-space nonlinearity as an equivalent image-domain convolution. This enables analytical derivation of the Jacobian matrix and the g-factor map—first such derivation for nonlinear k-space interpolation. We further validate our analysis via activation masking, closed-form g-factor computation, automatic differentiation, and Monte Carlo simulations. Contribution/Results: Our theory reveals that nonlinear parameters—e.g., the negative slope of leaky ReLU—govern a fundamental trade-off between noise suppression and loss of spatial resolution/contrast, analogous to Tikhonov regularization. Analytical predictions match empirical results with high fidelity; central artifacts are traced to activation autocorrelation structure. Evaluated on brain MRI data, the framework enables controllable balancing of noise robustness and image fidelity.

Technology Category

• Machine Learning: Active Learning
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Search and Optimization: Non-convex Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs

Abstract

Purpose: Noise resilience in image reconstructions by scan-specific robust artificial neural networks for k-space interpolation (RAKI) is linked to nonlinear activations in k-space. To gain a deeper understanding of this relationship, an image space formalism of RAKI is introduced for analyzing noise propagation analytically, identifying and characterizing image reconstruction features and to describe the role of nonlinear activations in a human readable manner. Methods: The image space formalism for RAKI inference is employed by expressing nonlinear activations in k-space as element-wise multiplications with activation masks, which transform into convolutions in image space. Jacobians of the de-aliased, coil-combined image relative to the aliased coil images can be expressed algebraically, and thus, the noise amplification is quantified analytically (g-factor maps). We analyze the role of nonlinearity for noise resilience by controlling the degree of nonlinearity in the reconstruction model via the negative slope parameter in leaky ReLU. Results: The analytical g-factor maps correspond with those obtained from Monte Carlo simulations and from an auto differentiation approach for in vivo brain images. Apparent blurring and contrast loss artifacts are identified as implications of enhanced noise resilience. These residual artifacts can be traded against noise resilience by adjusting the degree of nonlinearity in the model (Tikhonov-like regularization) in case of limited training data. The inspection of image space activations reveals an autocorrelation pattern leading to a potential center artifact. Conclusion: The image space formalism of RAKI provides the means for analytical quantitative noisepropagation analysis and human-readable visualization of the effects of the nonlinear activation functions in k-space.

Problem

• Analyzes noise resilience in RAKI reconstructions via image space formalism
• Quantifies noise amplification using g-factor maps and nonlinearity control
• Identifies artifacts and trade-offs between noise resilience and image quality

Innovation

• Image space formalism analyzes RAKI noise resilience
• Nonlinear activations transform into image convolutions
• Adjustable nonlinearity balances noise and artifacts