Next-Generation Phase Unwrapping for Dense MRI using Deep Learning

Publication ID: 24-11857288_0005_PTD
Published: November 07, 2025
Category:Future Evolutions & Paradigm Shifts

Legal Citation

pr1or.art Inc., “Next-Generation Phase Unwrapping for Dense MRI using Deep Learning,” Published Technical Disclosure No. 24-11857288_0005_PTD, Published November 07, 2025, available at https://archive.pr1or.art/24-11857288_0005_PTD
This technical disclosure describes improvements that would be readily apparent to a Person Having Ordinary Skill In The Art (PHOSITA) when considered in combination with the foundational architecture disclosed in U.S. Patent No. 11,857,288.

Summary of the Inventive Concept

A novel system and method for real-time phase unwrapping for dense MRI using deep learning, enabling accurate and efficient cardiac strain analysis.

Background and Problem Solved

The original patent, 'Systems and methods for phase unwrapping for dense MRI using deep learning', has limitations in terms of processing speed and accuracy. The new inventive concept addresses these limitations by introducing real-time phase unwrapping capabilities, improved spatial and temporal dependencies learning, and personalized cardiac strain analysis.

Detailed Description of the Inventive Concept

The new inventive concept comprises a neural network trained on a dataset of cardiac MRI images to predict phase unwrapping labels. The neural network is adapted to learn spatial and temporal dependencies in the cardiac MRI images, enabling accurate and efficient phase unwrapping. The system further includes a processing unit configured to receive the phase unwrapped MRI images and apply a deep learning model to generate a cardiac strain map. Additionally, the inventive concept enables personalized cardiac strain analysis using patient-specific cardiac strain analysis models.

Novelty and Inventive Step

The new claims introduce real-time phase unwrapping capabilities, improved spatial and temporal dependencies learning, and personalized cardiac strain analysis, which are not present in the original patent. The inventive concept's ability to learn spatial and temporal dependencies in cardiac MRI images and generate accurate phase unwrapped images in real-time constitutes a significant improvement over the original patent.

Alternative Embodiments and Variations

Alternative embodiments of the inventive concept could include the use of different deep learning architectures, such as transformers or graph neural networks, to improve the accuracy and efficiency of phase unwrapping. Additionally, the inventive concept could be adapted for use with other medical imaging modalities, such as CT or ultrasound.

Potential Commercial Applications and Market

The inventive concept has significant commercial potential in the medical imaging and diagnostics industry, particularly in the areas of cardiac strain analysis and personalized medicine. The ability to perform real-time phase unwrapping and generate accurate cardiac strain maps could revolutionize the field of cardiac imaging, enabling faster and more accurate diagnoses.

Field of Art

Medical imaging, specifically cardiac magnetic resonance imaging (MRI) with a focus on deep learning-based image processing and strain analysis techniques

Person of Ordinary Skill (PHOSITA) Profile

A skilled practitioner with expertise in medical image processing, deep learning neural networks, signal processing for MRI, and cardiac imaging analysis, typically holding a PhD in biomedical engineering, medical physics, or computer science with specialized training in machine learning and medical image analysis

Obviousness Rationale

A person having ordinary skill in the art would recognize that the published technical disclosure represents predictable extensions of the source patent's deep learning approach to phase unwrapping in cardiac MRI. The variations introduce incremental improvements in neural network architecture, real-time processing, and personalized analysis that would be considered routine optimization techniques within the field of medical image processing and deep learning. The core technical problem of phase unwrapping and strain analysis remains consistent, with the PTD offering straightforward algorithmic and architectural enhancements.

Obvious Combinations & Variations

Source Patent Element
U-Net structured CNN for computing wrapping label map
PTD Variation
Neural network adapted to learn spatial and temporal dependencies with potential alternative architectures like transformers or graph neural networks
Obviousness Reasoning
Exploring alternative neural network architectures for image processing is a known technique, with predictable outcomes of potentially improved performance through architectural variation
Source Patent Element
Displacement encoded MRI data for cardiac strain analysis
PTD Variation
Real-time phase unwrapping and generation of personalized cardiac strain maps using patient-specific models
Obviousness Reasoning
Extending computational techniques to provide more personalized and efficient processing represents a predictable evolution of existing medical imaging technologies
Source Patent Element
Deep learning models for phase unwrapping in cardiac MRI
PTD Variation
Applying deep learning models across different medical imaging modalities like CT or ultrasound
Obviousness Reasoning
Cross-modal application of image processing techniques is a standard approach in medical imaging, representing a logical and obvious extension of existing methodologies
Source Patent Element
CNN for epicardial and endocardial segmentation
PTD Variation
Enhanced neural network trained to learn spatial and temporal dependencies with improved segmentation capabilities
Obviousness Reasoning
Incremental improvements in neural network training and feature extraction are expected developments in deep learning image processing technologies
Source Patent Element
Phase unwrapping techniques for cardiac MRI
PTD Variation
Real-time processing and generation of cardiac strain maps using advanced deep learning techniques
Obviousness Reasoning
Improving computational efficiency and accuracy through advanced machine learning techniques represents a predictable progression in medical image processing technologies
35 U.S.C. § 103 Summary: Based on the teachings of US Patent 11857288 and the published technical disclosure, a person having ordinary skill in the art would find the claimed variations in deep learning-based phase unwrapping and cardiac strain analysis techniques to be obvious extensions of the prior art. The incremental improvements in neural network architecture, real-time processing, and personalized analysis represent routine optimization techniques that would be apparent to a skilled practitioner in medical image processing and machine learning.

Original Patent Information

Patent NumberUS 11,857,288
TitleSystems and methods for phase unwrapping for dense MRI using deep learning
Assignee(s)University of Virginia Patent Foundation