CogGen: Cognitive-Load-Informed Fully Unsupervised Deep Generative Modeling for Compressively Sampled MRI Reconstruction

2026-02-20

AI Summary

This work addresses the challenge of ill-posed inverse problems in compressed sensing MRI (CS-MRI) reconstruction under limited training data or computational resources, where conventional unsupervised generative models often suffer from slow convergence and overfitting to noise. To mitigate this, the authors propose a cognitive-load-guided self-paced curriculum learning framework that formulates reconstruction as a staged inversion process: it initially focuses on low-frequency, high signal-to-noise ratio k-space data and progressively incorporates high-frequency or noise-dominated measurements. Learning difficulty is dynamically modulated through a dual mechanism of soft weighting and hard selection. By integrating Deep Image Prior (DIP) and Implicit Neural Representations (INR), the framework yields two variants—CogGen-DIP and CogGen-INR—that significantly enhance reconstruction fidelity and convergence speed in unsupervised settings, achieving performance comparable to certain supervised methods.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Learning & Optimization for CV
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Fully unsupervised deep generative modeling (FU-DGM) is promising for compressively sampled MRI (CS-MRI) when training data or compute are limited. Classical FU-DGMs such as DIP and INR rely on architectural priors, but the ill-conditioned inverse problem often demands many iterations and easily overfits measurement noise. We propose CogGen, a cognitive-load-informed FU-DGM that casts CS-MRI as staged inversion and regulates task-side"cognitive load"by progressively scheduling intrinsic difficulty and extraneous interference. CogGen replaces uniform data fitting with an easy-to-hard k-space weighting/selection strategy: early iterations emphasize low-frequency, high-SNR, structure-dominant samples, while higher-frequency or noise-dominated measurements are introduced later. We realize this schedule through self-paced curriculum learning (SPCL) with complementary criteria: a student mode that reflects what the model can currently learn and a teacher mode that indicates what it should follow, supporting both soft weighting and hard selection. Experiments and analyses show that CogGen-DIP and CogGen-INR improve reconstruction fidelity and convergence behavior compared with strong unsupervised baselines and competitive supervised pipelines.

Problem

• compressive sensing MRI
• unsupervised deep generative modeling
• inverse problem
• overfitting
• reconstruction fidelity

Innovation

• cognitive-load-informed
• fully unsupervised deep generative modeling
• self-paced curriculum learning
• k-space weighting
• compressive MRI reconstruction