AI-Assisted PET Attenuation Correction for Artifact Mitigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing attenuation correction methods in Positron Emission Tomography (PET) data introduce artifacts due to tissue displacement between PET and CT acquisitions, leading to inaccurate quantitative measurements and lesion detection.
Innovation Solution
A system and method that generates CT-based attenuation correction data, predicts attenuation correction data based on non-attenuation corrected PET image data, and modifies CT-based attenuation data to reconstruct PET image data with reduced artifacts, using AI techniques for patient-specific correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If CT-based attenuation correction is used, then quantitative accuracy is improved, but artifacts are introduced due to tissue displacement between PET and CT acquisitions
Solution Approach 1:
The system uses a feedback loop where the attenuation correction artifact mitigator analyzes the PET image data for presence of attenuation correction artifact, predicts corrected attenuation data based on non-attenuation corrected PET image data, and updates the CT-based attenuation correction data. This iterative feedback process continuously refines the attenuation correction to eliminate artifacts while preserving quantitative accuracy.
Solution Approach 2:
The patent introduces an intermediary approach by using deep learning-based prediction models as a mediator between the CT-based attenuation correction data and the final PET image reconstruction. The inference engine predicts attenuation correction data from non-attenuation corrected PET image data, serving as an intermediary that bridges the discrepancy caused by tissue displacement between modalities.
2Object-affected harmful factors
If emission-based estimation of attenuation data is used, then artifacts are reduced, but processing time increases significantly
Solution Approach 1:
The system performs preliminary action by using deep learning models to pre-predict attenuation correction data from non-attenuation corrected PET image data during the reconstruction process. This allows the system to avoid time-consuming emission-based estimation methods while still achieving artifact reduction, as the AI model has pre-learned the relationship between PET signals and attenuation characteristics.
Solution Approach 2:
The patent replaces the computationally intensive mechanical process of emission-based attenuation estimation with an AI-based prediction system. The inference engine uses trained neural networks to substitute the complex iterative emission-based calculations, dramatically reducing processing time while maintaining accuracy in artifact mitigation.
3Productivity
If deep learning-based methods are used to generate attenuation data, then processing speed is improved, but new artifacts are introduced in the attenuation data
Solution Approach 1:
The system implements feedback by using the attenuation correction artifact mitigator to analyze the PET image data for the presence of attenuation correction artifact. This feedback mechanism allows the system to detect when deep learning-based attenuation data generation introduces artifacts and automatically adjust or refine the prediction to eliminate these new artifacts while maintaining processing speed benefits.
Solution Approach 2:
The patent applies parameter changes by dynamically adjusting the deep learning model's prediction based on the analyzed PET image characteristics. When artifacts are detected, the system modifies the attenuation correction data parameters through the attenuation correction data updater, refining the prediction to remove artifacts while preserving the computational efficiency of the deep learning approach.
4Reliability
If atlas-based registration methods are used, then patient-specific correction is achieved, but the process is time consuming and not fully patient specific
Solution Approach 1:
The system replaces the time-consuming mechanical process of atlas-based registration with an AI-based inference engine that directly predicts attenuation correction data from patient-specific non-attenuation corrected PET image data. This substitution eliminates the need for time-consuming atlas matching while achieving fully patient-specific correction through individualized deep learning predictions.
Solution Approach 2:
The patent uses copying by creating a digital copy of the patient's non-attenuation corrected PET image data as input to the deep learning model. The inference engine then generates a predicted attenuation correction data copy that is specifically tailored to the patient's anatomy and physiology, replacing the time-consuming process of copying and registering anatomical atlases.
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 diagnostic confidence and quantitative accuracy of PET data by mitigating attenuation correction artifacts, enhancing lesion detection and treatment monitoring.
Implementation Method 1
the radionuclide undergoes positron emission decay and emits a positron
Implementation Method 2
When the positron collides with an electron in the surrounding tissue, both the positron and the electron are annhilated and converted into a pair of photons
Implementation Method 3
When the photons impinge upon scintillation crystals of the detectors, a scintillation event (e.g., flash of light) is produced
Implementation Method 4
As the two photons travel along the LOR, they traverse tissue such as bone, muscle, soft tissue, etc., that attenuates the photons, resulting in a reduction in the detected signal
Data Source
AI summary
A system includes an attenuation corrector configured to generate Computed Tomography-(CT-) based attenuation correction data from CT image data, a Positron Emission Tomography (PET) reconstructor configured to reconstruct first PET image data based on PET projection data and the CT-based attenuation correction data, an attenuation correction artifact mitigator configured to analyze the first PET image data for a presence of attenuation correction artifact, an inference engine configured to predict attenuation correction data based on non-attenuation corrected PET image data in response to the presence of attenuation correction artifact in the first PET image data, and an attenuation correction data updater configured to generate modified attenuation correction data based on the CT-based attenuation correction data and the predicted attenuation correction data. The PET reconstructor is further configured to reconstruct second PET image data based on the PET projection data and the modified attenuation correction data.