Activity-Guided Dynamic PET Reconstruction for Fine-Structure Preservation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current PET imaging methods under-utilize dynamic imaging protocols, which could provide more accurate information than static imaging, and existing methods for analyzing dynamic PET data struggle with properly tuning penalty functions to suppress noise while preserving image features.

Innovation Solution

A method incorporating a spatially variant penalty function based on an activity map for dynamic PET reconstruction, using an objective function with edge-preserving potential and varying smoothing strength based on activity levels to generate parametric images.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Object-affected harmful factors

If a penalty function is applied to suppress noise in dynamic PET images, then noise is reduced, but image features such as fine structures and sharp edges are degraded

Engineering Contradiction:
ImprovenoiseVSAvoidimage features (fine structures and sharp edges)
Core Design Contradiction:
Object-affected harmful factorsVSManufacturing precision

Solution Approach 1:

The patent applies a spatially variant penalty function where the penalty strength varies across different regions of the image based on the activity map. Specifically, the penalty parameter β is set to be larger in regions with lower activity (to suppress noise more strongly) and smaller in regions with higher activity (to preserve edges and fine structures). This local adaptation allows the penalty function to suppress noise where needed while preserving image features in high-activity regions.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent dynamically adjusts the penalty parameter β based on the activity map generated from the PET data. The penalty function is not fixed but adapts to the specific characteristics of each image region. The activity map is used to compute spatially variant penalty parameters, making the regularization strength data-driven and adaptable to the actual tracer distribution patterns.

Inventive Principle:
Principle #15Dynamics

2Device complexity

If a fixed penalty function is used for Patlak fitting, then the processing is simple, but the performance is insufficient to fully exploit spatial-temporal information

Engineering Contradiction:
Improveprocessing complexityVSAvoidquantitative index accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent changes the penalty parameter from a fixed value to a spatially variant parameter that depends on the activity map. The penalty function transitions from a uniform regularization to a non-uniform regularization where β varies across the image based on local activity levels. This parameter change enables the system to adapt the regularization strength to the specific characteristics of each region, improving the exploitation of spatial-temporal information.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If dynamic PET imaging is implemented, then more accurate kinetic information is obtained, but the difficulty of properly tuning penalty parameters increases

Engineering Contradiction:
Improvekinetic properties characterizationVSAvoidpenalty function tuning complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent employs a self-service approach where the activity map, which is derived from the PET data itself, is used to determine the penalty parameters. Instead of requiring external tuning or manual adjustment of penalty parameters, the system automatically adapts the regularization strength based on the actual tracer distribution patterns in the image. This self-adjusting mechanism simplifies the tuning process while maintaining optimal performance.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent implements a feedback mechanism where the activity map (derived from the PET images) is fed back into the penalty function parameter determination. The activity map provides information about the tracer distribution, and this information is used to adjust the penalty parameters accordingly. This feedback loop enables the system to automatically optimize the balance between noise suppression and feature preservation based on the actual image content.

Inventive Principle:
Principle #23Feedback

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

Improves image quality by suppressing noise and preserving fine structures in PET images, allowing for more accurate characterization of tracer distribution and kinetic properties.

Implementation Method 1

When an emitted positron collides with an electron, an annihilation event occurs, wherein the positron and electron are combined. Most of the time, an annihilation event produces two gamma rays (at 511 keV) traveling at substantially 180 degrees apart.

Methodology Applied
Scientific EffectPositron-electron annihilation:

Implementation Method 2

By detecting the two gamma rays, and drawing a line between their locations, i.e., the line-of-response (LOR), one can determine the likely location of the original disintegration.

Methodology Applied
Scientific EffectGamma ray detection:

Implementation Method 3

a method including an optimal penalty function for dynamic PET that is able to exploit spatial information embed in dynamic PET data to suppress noise while preserving image features

Methodology Applied
Scientific EffectNoise suppression through penalty function:

Data Source

PatentEP3901917B1Medical image processing apparatus and medical image processing method
Publication Date: 2025.09.17 CANON MEDICAL SYST CORP
  • EP3901917B1 patent drawingFigure 1
  • EP3901917B1 patent drawingFigure 2
  • EP3901917B1 patent drawingFigure 3A

AI summary

A medical image processing apparatus according to an embodiment includes processing circuitry. The processing circuitry (370) obtains a plurality of dynamic sinograms, each of the plurality of dynamic sinograms representing detection events of gamma rays at a plurality of scintillator elements, sums the plurality of dynamic sinograms to generate an activity map based on a radioactivity level of the gamma rays, and generates, using the plurality of dynamic sinograms and the activity map, at least one parametric image.