Physiological Motion Field Prediction for Accurate ROI Measurement

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing medical imaging systems struggle with accurately measuring physiological motion during scans, leading to inaccuracies in disease diagnosis and treatment due to the generation of images corresponding to different motion phases of the region of interest.

Innovation Solution

A system utilizing a motion prediction model trained through supervised or unsupervised learning techniques to determine the motion field between reference and target images of an ROI, enabling accurate physiological condition assessment without relying on prior shape models.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional image processing methods are used to measure physiological motion, then the measurement process can be completed, but the measurement precision is insufficient and reliability is low

Engineering Contradiction:

Improvephysiological motion measurement precisionVSAvoidmeasurement reliability

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The patent replaces traditional mechanical image processing methods with a machine learning-based motion prediction model. The model takes reference and target images as input and directly predicts the motion field through trained neural network layers, eliminating the need for complex traditional image registration and feature tracking algorithms, thereby significantly improving measurement precision and reliability

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent transforms the motion measurement approach by changing from manual feature extraction and traditional image registration parameters to machine learning-based motion field prediction. The system uses trained model parameters (weights and biases) to directly compute motion vectors and deformation fields, enabling more accurate and reliable physiological motion measurement

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If annotated training samples are used to train the motion prediction model, then the model accuracy can be improved, but the data preparation complexity and time consumption increase

Engineering Contradiction:

Improvemotion field prediction accuracyVSAvoiddata preparation complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent implements self-service by enabling the system to automatically generate and annotate training data without human intervention. The data preparation module automatically extracts feature points, calculates motion fields, and generates annotated training samples from medical images, allowing the system to train itself without requiring manual annotation by experts, thus reducing complexity and time consumption

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent applies preliminary action by pre-processing and preparing training data automatically before model training. The data preparation module performs preliminary tasks including feature point extraction, motion field calculation, and annotation generation in advance, so that when the model is trained, the annotated data is already ready, significantly reducing the overall complexity and time required

Inventive Principle:

Principle #10Preliminary action

3Productivity

If manual feature extraction and traditional image registration are used, then the measurement process can be completed, but the productivity is low and time-consuming

Engineering Contradiction:

Improvemotion measurement productivityVSAvoidmeasurement time

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The patent replaces time-consuming manual feature extraction and traditional image registration processes with a machine learning-based motion prediction model. The model directly predicts motion fields from image pairs through trained neural network layers, eliminating iterative optimization and manual intervention, thereby dramatically improving productivity and reducing measurement time

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent uses copying by creating a trained motion prediction model that captures the essential motion patterns from training data. Once trained, the model can rapidly predict motion fields for new image pairs by simply copying the learned patterns, avoiding the need to reperform complex feature extraction and registration calculations, thus significantly increasing productivity

Inventive Principle:

Principle #26Copying

Data Source

PatentUS12518404B2Systems and methods for machine learning based physiological motion measurement

Publication Date: 2026.01.06 SHANGHAI UNITED IMAGING INTELLIGENCE CO LTD

AI summary

A system for physiological motion measurement is provided. The system may acquire a reference image corresponding to a reference motion phase of an ROI and a target image of the ROI corresponding to a target motion phase, wherein the reference motion phase may be different from the target motion phase. The system may identify one or more feature points relating to the ROI from the reference image, and determine a motion field of the feature points from the reference motion phase to the target motion phase using a motion prediction model. An input of the motion prediction model may include at least the reference image and the target image. The system may further determine a physiological condition of the ROI based on the motion field.