AI-Based MR Reconstruction Offloading to Remote Servers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
MR imaging systems face challenges with high memory and computation power usage due to AI reconstruction processes, which are not easily updated to keep pace with rapidly evolving AI solutions, given their limited hardware capabilities.
Innovation Solution
Offloading AI-based reconstruction to remote servers with greater processing resources, allowing for load balancing and deployment of various AI reconstruction processes, including machine-learned models, to handle k-space data and improve performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If AI reconstruction processes are deployed on the MR imaging system, then image quality is improved, but memory and computation power usage increase beyond the limited hardware capabilities
Solution Approach 1:
The patent extracts the AI reconstruction process from the MR imaging system and relocates it to a remote server. The scanner only performs data acquisition and transmits raw data, while the computationally intensive AI reconstruction is performed remotely, thereby reducing the computational burden on the scanner's limited hardware while maintaining high image quality.
Solution Approach 2:
The patent introduces a remote server as an intermediary between the MR scanner and the reconstruction process. This intermediary handles the computationally intensive AI reconstruction tasks, allowing the scanner to maintain its limited hardware capabilities while still benefiting from advanced AI-based image reconstruction.
2Productivity
If the MR imaging system is updated with new AI solutions, then reconstruction performance is improved, but the system cannot be updated rapidly due to hardware limitations
Solution Approach 1:
By extracting the AI reconstruction software from the scanner hardware, the system allows independent updates of AI models on the remote server without requiring hardware modifications or system reconfiguration at the scanner site. This enables rapid deployment of new AI solutions.
Solution Approach 2:
The remote server serves as a universal platform that can host multiple different AI reconstruction models and algorithms. This universal infrastructure can rapidly adapt to new AI solutions by simply updating software on the server, making the system highly adaptable without changing the scanner hardware.
3Adaptability or versatility
If multiple AI models are deployed for different reconstruction needs, then versatility is improved, but device complexity increases
Solution Approach 1:
The patent merges multiple AI models and reconstruction algorithms into a single remote server environment. This consolidation allows the system to support diverse reconstruction needs through one unified platform, avoiding the complexity of maintaining multiple separate systems or updating the scanner for each new model.
Solution Approach 2:
The scanner simply copies and transmits raw data to the remote server, which then applies the appropriate AI model. This copying approach allows the same hardware to work with multiple different AI models without modification, as the model selection and execution happen remotely on the server.
Data Source
AI summary
For magnetic resonance (MR) reconstruction using artificial intelligence (AI), the AI-based reconstruction for MR imaging systems is offloaded to one or more servers. A remote server performs AI-based reconstruction. A library of recent, old, custom, and/or publicly available AI-based reconstruction processes may be rapidly deployed and available to the server, which has the memory and processing resources for AI-based reconstruction. Load balancing of the data and/or between servers may improve performance.