Hybrid Factorized System Model for TOF PET Image Reconstruction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current list-mode time-of-flight (TOF) PET systems face limitations in image reconstruction accuracy due to the constraints of on-the-fly calculation of lines of response (LORs), which affects the spatial resolution and noise recovery in reconstructed images, especially when using factorized system models that do not account for detector response and photon non-colinearity.

Innovation Solution

A hybrid factorized system model combining a geometrical projection matrix with an image blurring matrix, where the image blurring kernel is estimated based on both analytical and measurement-based data, allowing for accurate modeling of spatially variant response without requiring detector blurring operations, and using a simple line integral-based LOR model for fast on-the-fly geometrical projection.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If an accurate system model is used for TOF PET image reconstruction, then image quality and spatial resolution are improved, but computational overhead and reconstruction time increase significantly

Engineering Contradiction:

Improveimage reconstruction accuracyVSAvoidreconstruction speed

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The system model is segmented into a geometrical projection matrix and an image blurring matrix. The geometrical projection matrix handles the mapping from image space to projection space with simple pure geometrical LORs, while the image blurring matrix separately models the detector physical response including point spread function effects. This segmentation allows each component to be optimized independently, reducing overall computational overhead while maintaining reconstruction accuracy.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The image blurring matrix incorporates spatially variant point spread function kernels that are tailored to local characteristics of the imaging geometry and detector response. Each voxel or region can have its own customized blurring kernel, allowing accurate modeling of local effects without requiring a full accurate system model for every calculation, thus improving efficiency while maintaining local reconstruction quality.

Inventive Principle:

Principle #3Local quality

2Productivity

If a factorized system model with detector blurring matrix is used, then computational efficiency is improved, but compatibility with list mode data is lost due to requirement of neighboring projection bins

Engineering Contradiction:

Improvecomputational efficiencyVSAvoidcompatibility with list mode data

Core Design Contradiction:

ProductivityVSAdaptability or versatility

Solution Approach 1:

Instead of applying detector blurring in projection space (which requires neighboring bins and is incompatible with list mode), the approach inverts the conventional factorization by placing the blurring operation in image space. The image blurring matrix operates on reconstructed image data rather than projection data, making it compatible with list mode reconstruction while maintaining computational efficiency through the factorized structure.

Inventive Principle:

Principle #13The other way round (Inversion)

3Speed

If simple geometrical LORs are used in projection matrix, then calculation speed is improved, but modeling accuracy of detector physical response deteriorates

Engineering Contradiction:

Improvecalculation speedVSAvoiddetector response modeling accuracy

Core Design Contradiction:

SpeedVSMeasurement precision

Solution Approach 1:

The image blurring matrix acts as an intermediary that compensates for the simplified geometrical LORs. While the projection matrix uses computationally efficient simple geometrical LORs without detailed detector physics, the image blurring matrix introduces appropriate point spread function kernels that restore the missing detector physical response effects, achieving both speed and accuracy.

Inventive Principle:

Principle #24Intermediary (Mediator)

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 enhances the accuracy of TOF PET image reconstruction by improving spatial resolution and lesion contrast recovery while reducing computational overhead, achieving comparable results to full system matrix modeling with faster reconstruction speeds.

Implementation Method 1

As the radiopharmaceutical decays, positrons are generated

Methodology Applied

Scientific EffectRadioactive decay: Radioactive Decay

Implementation Method 2

each of a plurality of positrons reacts with an electron in what is known as a positron annihilation event, thereby generating a pair of coincident gamma photons

Methodology Applied

Scientific EffectPositron annihilation:

Implementation Method 3

the time within the coincidence interval at which each gamma photon in the coincident pair is detected is also measured. The time-of-flight information provides an indication of the annihilation location

Methodology Applied

Scientific EffectTime of flight: Time of Flight

Data Source

PatentUS10216701B2Image-based point-spread-function modelling in time-of-flight positron-emission-tomography iterative list-mode reconstruction

Publication Date: 2019.02.26 CANON KK

AI summary

A method of calculating a system matrix for time-of-flight (TOF) list-mode reconstruction of positron-emission tomography (PET) images, the method including determining a TOF geometric projection matrix G including effects of object attenuation; estimating an image-blurring matrix R in image space; obtaining a diagonal matrix D that includes TOF-based normalization factor; and calculating the system matrix H as H=DGR.