Deep Learning TOF Models for BGO Cherenkov PET/CT Imaging

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Modern PET systems face challenges in achieving high-quality TOF data without the high cost and reduced design flexibility associated with using lutetium oxyorthosilicate (LSO) crystals, and lower-quality TOF data from BGO crystals results in suboptimal image reconstruction due to lower-quality Cherenkov luminescence.

Innovation Solution

A method involving a deep learning neural network model that enhances TOF data quality by combining real Cherenkov luminescence-based data with lower-quality TOF data, using BGO crystals, to improve image quality through a convolutional neural network (CNN) model.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If LSO crystals are used to achieve high-quality TOF data, then image quality and TOF performance are improved, but system cost increases significantly

Engineering Contradiction:

ImproveTOF data qualityVSAvoidsystem cost

Core Design Contradiction:

Measurement precisionVSEase of manufacture

Solution Approach 1:

The patent uses BGO crystals, which are cheaper than LSO crystals, to build the PET detector system. While BGO crystals produce lower quality TOF data through Cherenkov luminescence compared to LSO's scintillation, the patent compensates for this limitation through advanced image reconstruction algorithms and processing techniques that enhance the utility of the lower-quality TOF data, thereby achieving cost reduction while maintaining acceptable performance

Inventive Principle:

Principle #27Cheap short-living objects (Disposable)

Solution Approach 2:

The patent changes the operational parameters and processing methods to optimize performance with BGO crystals. This includes using specific reconstruction algorithms, adjusting energy windows, and implementing advanced TOF data processing techniques to maximize the quality of TOF images derived from Cherenkov luminescence in BGO, thereby achieving good performance without the higher cost of LSO crystals

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If LSO crystals are used to achieve high-quality TOF data, then TOF performance is improved, but design flexibility is reduced

Engineering Contradiction:

ImproveTOF performanceVSAvoiddesign flexibility

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent employs BGO crystals that can serve multiple functions: they provide stopping power for 511 keV photons, generate Cherenkov luminescence for TOF measurements, and allow for various detector design configurations. This multi-functionality maintains design flexibility while achieving acceptable TOF performance through alternative mechanisms compared to LSO crystals

Inventive Principle:

Principle #6Universality (Multi-functionality)

3Ease of manufacture

If BGO crystals are used to reduce cost, then system cost decreases, but TOF data quality deteriorates

Engineering Contradiction:

Improvesystem costVSAvoidTOF data quality

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent uses Cherenkov luminescence as an intermediary mechanism in BGO crystals to generate TOF signals. While the Cherenkov light yield is lower than scintillation in LSO, the patent employs specialized photodetectors and signal processing techniques to effectively detect and utilize this Cherenkov light, thereby achieving acceptable TOF data quality from the cheaper BGO crystals

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent substitutes the scintillation mechanism of LSO with the Cherenkov radiation mechanism in BGO crystals. This substitution changes the physical mechanism for generating TOF signals from scintillation photons to Cherenkov photons, requiring different detection and processing approaches but enabling the use of cheaper BGO material while maintaining TOF functionality

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

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

Enhances image quality by increasing the Signal to Noise Ratio (SNR) in PET images, achieving comparable results to systems using LSO crystals while reducing costs and maintaining design flexibility.

Implementation Method 1

some TOF data may be generated (with high TOF performance) by exploiting Cherenkov luminescence in the BGO crystals

Methodology Applied

Scientific EffectCherenkov luminescence: Cherenkov Effect

Implementation Method 2

enhances image quality by increasing the Signal to Noise Ratio (SNR) in PET images

Methodology Applied

Scientific EffectSignal to Noise Ratio enhancement:

Data Source

PatentUS20250278818A1Deep learning TOF models for cherenkov based pet/ct

Publication Date: 2025.09.04 GE PRECISION HEALTHCARE LLC

AI summary

The current disclosure provides systems and methods for increasing a quality of Positron Emission Tomography (PET) images generated by a PET system with detectors including bismuth germinate (BGO) crystals. In one example, a method for the PET system comprises extracting a higher-quality 2-D image from a first image volume reconstructed using higher-quality time-of-flight (TOF) data acquired from a subject during a scan performed using the PET system, the TOF data based on Cherenkov radiation detected at the BGO crystals; extracting a lower-quality 2-D image from a second image volume reconstructed using lower-quality TOF data acquired from the subject by the PET system; generating an enhanced-quality 2-D image from the lower-quality 2-D image using a trained image quality enhancement model; merging the enhanced-quality 2-D image with the higher-quality 2-D image; and displaying the merged 2-D image on a display device of the PET system.