PET Parametric Image Reconstruction Using Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing PET image reconstruction methods using indirection algorithms introduce noise and degrade image quality, while direct reconstruction algorithms are complex and slow, necessitating improved systems and methods for reconstructing parametric images with enhanced quality and speed.
Innovation Solution
A system utilizing a target machine learning model, such as a deep learning neural network, to map image sequences to parametric images, incorporating training and validation data to improve reconstruction speed and quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If indirect reconstruction algorithm is used, then reconstruction speed is improved, but image quality deteriorates due to increased noise
Solution Approach 1:
The patent introduces a machine learning model as an intermediary between the indirect reconstruction algorithm and the final parametric image. The model takes the preliminary reconstructed images as input and processes them to produce high-quality parametric images, effectively mediating between speed and quality requirements
Solution Approach 2:
The patent replaces the traditional mechanical iterative reconstruction process with a machine learning-based system. The neural network model learns the mapping from preliminary images to parametric images, substituting the complex iterative mathematical operations with a trained computational model that achieves both speed and quality
2Measurement precision
If direct reconstruction algorithm is used, then image quality is improved, but device complexity increases and reconstruction speed decreases
Solution Approach 1:
The patent segments the reconstruction process into two distinct stages: first, an indirect reconstruction algorithm generates preliminary images quickly; second, a machine learning model processes these preliminary images to produce the final high-quality parametric images. This segmentation divides the complex task into manageable parts with different optimization goals
Solution Approach 2:
The patent performs preliminary reconstruction using a simplified indirect algorithm before applying the machine learning model. This preliminary action creates a starting point that the subsequent model can refine, avoiding the need to perform the entire complex direct reconstruction from scratch
3Measurement precision
If direct reconstruction algorithm is used, then image quality is improved, but reconstruction time increases
Solution Approach 1:
The patent maintains continuous useful action by processing data through multiple stages without idle periods. The indirect reconstruction runs continuously to generate preliminary images, which immediately feed into the machine learning model for final processing, ensuring that computational resources are continuously productive throughout the pipeline
Data Source
AI summary
Methods and systems for PET image reconstruction are provided. A method may include obtaining an image sequence associated with a subject. The image sequence may include one or more images generated via scanning the subject at one or more consecutive time periods. The method may also include obtaining a target machine learning model. The method may further include generating at least one target image using the target machine learning model based on the image sequence. The at least one target image may present a dynamic parameter associated with the subject. The target machine learning model may provide a mapping between the image sequence and the at least one target image.