Scattering Estimation in PET Imaging Using Adaptive Convolution Kernels

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing scattering estimation methods for positron emission tomography (PET) images face issues with high calculation costs and errors due to empirically determined convolution function parameters, leading to noise and image quality degradation during scattering correction.

Innovation Solution

A method that determines a convolution kernel based on the scattered radiation index value from PET measurement data and absorption coefficient data, allowing for the simulation of multiple scattering distributions without direct estimation, using a single scattering distribution and a weighted average filter, thereby reducing processing time and improving accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If convolution kernel parameters are determined empirically, then the scattering estimation method is simple to implement, but errors occur between estimation results and actual measurement values, causing noise or voids in obtained images

Engineering Contradiction:

ImproveEase of implementationVSAvoidAccuracy of scattering estimation

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent changes the approach from empirically determining convolution kernel parameters to determining them based on scattered radiation index values calculated from actual measurement data. This parameter change allows the system to adapt to different imaging conditions and subject types, significantly improving accuracy while maintaining implementation simplicity through automated calculation.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The system performs self-service by automatically calculating the scattered radiation index value from the measurement data itself, eliminating the need for external empirical parameter selection. The convolution kernel parameters are derived directly from the data being processed, making the system self-adapting and removing dependency on manual parameter tuning.

Inventive Principle:

Principle #25Self-service

2Reliability

If multiple scattering distribution is directly estimated using convolution kernel fitting, then scattering correction can be performed, but the calculation cost is high and the time required for scattering estimation becomes longer

Engineering Contradiction:

ImproveScattering correction capabilityVSAvoidScattering estimation speed

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The patent segments the scattering estimation process into two parts: first calculating the scattered radiation index value from measurement data, then using this index to determine convolution kernel parameters for smoothing the single scattering distribution. This segmentation avoids the computationally intensive direct estimation of multiple scattering distribution while maintaining correction effectiveness.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system performs preliminary action by calculating the scattered radiation index value before performing the actual scattering correction. This pre-calculation step provides the necessary parameters for efficient convolution kernel application, avoiding the need for time-consuming iterative fitting processes during the main correction phase.

Inventive Principle:

Principle #10Preliminary action

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

This approach reduces the time required for scattering estimation and prevents image quality degradation by using a convolution kernel determined by actual measurement data, enhancing the accuracy of scattering correction.

Implementation Method 1

determining a convolution kernel for smoothing the single scattering distribution based on a scattered radiation index value of the radioactive image, and fitting, to the positron emission tomography measurement data, a scattering distribution smoothed by applying the convolution kernel to the single scattering distribution

Methodology Applied

Scientific EffectConvolution:

Data Source

PatentUS11002866B2Scattering estimation method and image processor

Publication Date: 2021.05.11 SHIMADZU CORP

AI summary

A scattering estimation method includes determining a convolution kernel for smoothing a single scattering distribution based on a scattered radiation index value (R) of a radioactive image (5) (S4) and fitting, to positron emission tomography measurement data, a scattering distribution smoothed by applying the convolution kernel to the single scattering distribution (S5).