AI Engine for Tomographic Image Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computerized tomography (CT) imaging systems often produce reconstructed volume data with artifacts, leading to image degradation and affecting subsequent diagnosis and radiotherapy treatment planning.
Innovation Solution
The use of AI engines with multiple processing layers and interposing back-projection or forward-projection modules to process 2D and 3D projection data, learning weight data during training phases to perform tomographic image reconstruction and analysis, thereby reducing artifacts and enhancing image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional CT reconstruction methods are used, then the imaging process is simple and fast, but artifacts are introduced and image quality deteriorates
Solution Approach 1:
The patent introduces an intermediary AI engine between the conventional CT reconstruction pipeline and the final image output. This AI engine processes projection data through multiple processing layers (including back-projection modules) to generate improved 3D volume data with reduced artifacts, while maintaining the overall simplicity of the CT imaging workflow.
Solution Approach 2:
The patent replaces traditional mechanical/conventional reconstruction algorithms with an AI-based processing system. The AI engine uses neural network layers to transform 2D projection data into 3D volume data, substituting the need for complex iterative reconstruction algorithms and achieving superior image quality with reduced computational complexity.
2Manufacturing precision
If AI engines with multiple processing layers are used, then artifact reduction and image quality improvement are achieved, but processing time and computational complexity increase
Solution Approach 1:
The AI engine performs preliminary processing of projection data through multiple layers during the reconstruction phase, preparing enhanced 3D volume data that reduces artifacts before subsequent analysis steps. This preliminary action ensures high image quality is achieved early in the processing pipeline, avoiding the need for time-consuming post-processing corrections.
Solution Approach 2:
The patent changes the processing parameters by using AI-based transformation instead of conventional iterative reconstruction. The AI engine processes data through multiple layers with optimized parameters to achieve fast convergence, reducing processing time while maintaining high reconstruction accuracy and artifact reduction.
3Reliability
If AI engines with multiple processing layers are used, then feature loss is reduced and image quality is enhanced, but system complexity and training requirements increase
Solution Approach 1:
The AI engine is segmented into multiple processing layers (including back-projection modules) that each perform specific functions in the reconstruction pipeline. This segmentation allows the complex task of artifact reduction and feature preservation to be divided into manageable stages, improving reliability while keeping the overall structure organized and maintainable.
Solution Approach 2:
The AI engine is designed with multi-functionality, where the same processing layers can handle various tasks including artifact reduction, feature enhancement, and data transformation. This universality reduces the need for separate specialized modules, thereby reducing overall system complexity while maintaining high feature preservation accuracy.
Data Source
AI summary
Example methods and systems for tomographic image reconstruction are provided. One example method may comprise: obtaining two-dimensional (2D) projection data and processing the 2D projection data using the AI engine that includes multiple first processing layers, an interposing back-projection module and multiple second processing layers. Example processing using the AI engine may involve: generating 2D feature data by processing the 2D projection data using the multiple first processing layers, reconstructing first three-dimensional (3D) feature volume data from the 2D feature data using the back-projection module; and generating second 3D feature volume data by processing the first 3D feature volume data using the multiple second processing layers.